Paperid:1

Authors:Aaditya Ramdas

Title: ICML Tutorial Game-theoretic Statistics and Sequential Anytime-Valid Inference

Abstract:
Sequential anytimevalid inference (SAVI) provides measures of statistical evidence and uncertainty --- e-values and e-processes for testing and confidence sequences for estimation --- that remain valid at all stopping times. These allow for continuous monitoring and analysis of accumulating data and optional stopping for any reason. These methods crucially rely on nonnegative martingales, which are wealth processes of a player in a betting game, thus yielding the area of "game-theoretic statistics". This tutorial will present the game-theoretic philosophy, intuition, language and mathematics behind SAVI, summarized in a recent a new book https://arxiv.org/pdf/2410.23614, to be published before ICML as the first edition of the new book series, Foundations and Trends in Statistics.

Paperid:2

Authors:Mark Tygert

Title: ICML Tutorial Calibration and Bias in Algorithms, Data, and Models: a tutorial on metrics and plots for measuring calibration, bias, fairness, reliability, and robustness

Abstract:
There are many different notions of bias and fairness. When comparing subpopulations, an especially important dichotomy is between (1) equal or equitable average outcomes and (2) equal or equitable treatment. In the particular context considered here, "equal treatment" and "equal opportunity" are not too different. However, comparing the average outcome of one subpopulation to another is different and sometimes less desirable than comparing the outcomes of pairs of individuals (one individual from each subpopulation) for which the individuals in each pair are similar. The latter requires comparing outcomes via "conditioning on" or "controlling for" confounding covariates.

Paperid:3

Authors:Parikshit Ram · Benjamin Hoover · Dmitry Krotov

Title: ICML Tutorial Modern Methods in Associative Memory

Abstract:
Associative Memories like the famous Hopfield Networks are elegant models for describing fully recurrent neural networks whose fundamental job is to store and retrieve information. In the past few years they experienced a surge of interest due to novel theoretical results pertaining to their information storage capabilities, and their relationship with SOTA AI architectures, such as Transformers and Diffusion Models. These connections open up possibilities for interpreting the computation of traditional AI networks through the theoretical lens of Associative Memories. Additionally, novel Lagrangian formulations of these networks make it possible to design powerful distributed models that learn useful representations and inform the design of novel architectures. This tutorial provides an approachable introduction to Associative Memories, emphasizing the modern language and methods used in this area of research, with practical handson mathematical derivations and coding notebooks.

Paperid:4

Authors:Mingzhi Wang · Chengdong Ma · Yaodong Yang

Title: ICML Tutorial Alignment Methods for Large Language Models

Abstract:
Large Language Model (LLM) alignment has become an increasingly critical topic in contemporary AI research, especially as LLMs continue to scale and integrate into realworld applications. Ensuring that LLMs generate outputs aligned with human values, preferences, and ethical considerations is essential for their safe and effective deployment. This tutorial aims to provide a comprehensive introduction to LLM alignment methods, offering a structured and accessible entry point for researchers and practitioners interested in the field. It will present key concepts and challenges, introduce fundamental approaches such as Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO), and, building on these foundations, review a spectrum of refinements and variants. In addition, it will cover recent advancements in game-theoretical approach to alignment and theoretical frameworks that provide a deeper understanding of alignment methodologies. Beyond theoretical insights, the tutorial will emphasize the practical aspects of LLM alignment, illustrating how these techniques are applied in real-world scenarios and guiding participants in building intuition about alignment strategies. By the end of the tutorial, attendees will gain a solid foundation in LLM alignment, equipping them with the knowledge needed to critically engage with the field, understand current research trends, and explore future directions.

Paperid:5

Authors:Amy Zhang · Benjamin Eysenbach

Title: ICML Tutorial Generative AI Meets Reinforcement Learning

Abstract:
This tutorial explores the intersection of generative AI and reinforcement learning, demonstrating how generative models can be understood as RL agents and environments, and conversely, how RL can be viewed as generative modeling. It aims to bridge the gap between these fields, showing how insights from each can enhance the other. The workshop will cover topics such as reinterpreting generative AI training through an RL lens, adapting generative AI to build new RL algorithms, and understanding how AI agents interacting with tools and humans create a new generative model. It will also discuss future directions and open problems, focusing on how RL can shape the future of foundation model training and enable generative AI systems to construct their own knowledge.

Paperid:6

Authors:Ziyu Yao · Daking Rai

Title: ICML Tutorial Tutorial on Mechanistic Interpretability for Language Models

Abstract:
Mechanistic interpretability (MI) is an emerging subfield of interpretability that seeks to understand a neural network model by reverse-engineering its internal computations. Recently, MI has garnered significant attention for interpreting transformer-based language models (LMs), resulting in many novel insights yet introducing new challenges. Given how fast this topic is now attracting the ML/AI community's attention, the goal of this tutorial is to provide a comprehensive overview of MI for LMs, including its historical contexts, the various techniques to implement and evaluate MI, findings and applications based on MI, and future challenges. The tutorial will particularly be presented following an innovative Beginner's Roadmap that the presenters carefully curated, aiming to enable researchers new to MI to quickly pick up this field and leverage MI techniques in their LM applications.

Paperid:7

Authors:Qing Qu · Yuxin Chen · Liyue Shen

Title: ICML Tutorial Harnessing Low Dimensionality in Diffusion Models: From Theory to Practice

Abstract:
Diffusion models have recently gained attention as a powerful class of deep generative models, achieving stateof-the-art results in data generation tasks. In a nutshell, they are designed to learn an unknown data distribution starting from Gaussian noise, mimicking the process of non-equilibrium thermodynamic diffusion. Despite their outstanding empirical successes, the mathematical and algorithmic foundations of diffusion models remain far from mature. For instance: (i) Generalization: it remains unclear how diffusion models, trained on finite samples, can generate new and meaningful data that differ from the training set; (ii) Efficiency: due to the enormous model capacity and the requirement of many sampling steps, they often suffer from slow training and sampling speeds; (iii) Controllability: it remains computationally challenging and unclear how to guide and control the content generated by diffusion models, raising challenges over controllability and safety, as well as solving inverse problems across many scientific imaging applications.

Paperid:8

Authors:Natalia Ponomareva · Sergei Vassilvitskii · Peter Kairouz · Alex Bie

Title: ICML Tutorial DP-fy your DATA: How to (and why) synthesize Differentially Private Synthetic Data

Abstract:
This tutorial focuses on the increasingly important area of differentially private (DP) synthetic data generation, addressing the need for robust anonymization in machine learning. Creating DP synthetic data allows for data sharing and analysis without compromising individuals' privacy, opening up possibilities for collaborative research and model training. The tutorial aims to bridge the gap between various related fields, such as DP training, DP inference, and empirical privacy testing, providing a comprehensive guide for generating DP synthetic data across different data types.

Paperid:9

Authors:Hamed Hassani · Amin Karbasi · Alexander Robey · Yaron Singer

Title: ICML Tutorial Jailbreaking LLMs and Agentic Systems: Attacks, Defenses, and Evaluations

Abstract:
Since the inception of language models (LLMs), considerable attention has been directed toward the field of AI safety. These efforts aim to identify a range of best practices—including evaluation protocols, defense algorithms, and content filters—that facilitate the ethical, trustworthy, and reliable deployment of LLMs and related technologies. A key component of AI safety is model alignment, a broad concept referring to algorithms that optimize the outputs of LLMs to align with human values. And yet, despite these efforts, recent research has identified several failure modes—referred to as jailbreaks—that circumvent LLM alignment by eliciting unsafe content from a targeted model. And while initial jailbreaks targeted the generation of harmful information (e.g., copyrighted or illegal material), modern attacks seek to elicit domainspecific harms, such as digital agents violating user privacy and LLM-controlled robots performing harmful actions in the physical world. In the worst case, future attacks may target self-replication or power seeking behaviors. The insidious nature of jailbreaking attacks represents a substantial obstacle to the broad adoption of LLMs. Therefore, it is critical for the machine learning community to study these failure modes and develop effective defense strategies that counteract them.

Paperid:10

Authors:qiang liu

Title: ICML Tutorial Flowing Through Continuous-Time Generative Models: A Clear and Systematic Tour

Abstract:
Continuoustime generative models—particularly diffusion- and flow-based models—have emerged as a dominant paradigm in generative AI, with applications in image, video, molecular, and audio synthesis, as well as scientific modeling. Despite their success, the field’s rich mathematical structure, varied terminology, and subtle theoretical foundations often lead to confusion and fragmented understanding.

Paperid:11

Authors:Leena Chennuru Vankadara

Title: ICML Tutorial Training Neural Networks at Any Scale

Abstract:
At the heart of deep learning’s transformative impact lies the concept of scaleencompassing both data and computational resources, as well as their interaction with neural network architectures.

Paperid:12

Authors:Jiaoda Li

Title: ICML Tutorial The Underlying Logic of Language Models

Abstract:
The formal basis of the theory of computation lies in the study of languages, subsets of Σ, the set of all strings over an alphabet Σ. Models of computation can be taxonomized into the languages they can decide on, i.e., which languages a model can be used to determine membership of. For instance, finitestate automata can decide membership in the regular languages. Language models are probabilistic generalizations of language where the notion of a set is relaxed into one of a probability distribution over Σ. Recently, language models parameterized using recurrent neural networks, transformers, and state-space models have achieved enormous success in natural language processing. Similarly to how theorists have taxonomized models of deterministic computation, researchers have been made to taxonomize the expressivity of language models based on various architectures in terms of the distributions over strings they can represent. This tutorial presents a self-contained overview of the formal methods used to taxonomize the expressivity of language models, which encompass formal language and automata theory, various forms of formal logic, circuit complexity, and programming languages such as RASP. For example, we illustrate how transformers, under varying assumptions, can be characterized by different fragments of formal logic.

Paperid:13

Authors:Sarah Meiklejohn · Hayden Blauzvern · Mihai Maruseac · Spencer Schrock · Laurent Simon · Ilia Shumailov

Title: Position: Machine Learning Models Have a Supply Chain Problem

Abstract:
Powerful machine learning (ML) models are now readily available online, which creates exciting possibilities for users who lack the deep technical expertise or substantial computing resources needed to develop them. On the other hand, this type of open ecosystem comes with many risks. In this paper, we argue that the current ecosystem for open ML models contains significant supplychain risks, some of which have been exploited already in real attacks. These include an attacker replacing a model with something malicious (e.g., malware), or a model being trained using a vulnerable version of a framework or on restricted or poisoned data. We then explore how Sigstore, a solution designed to bring transparency to open-source software supply chains, can be used to bring transparency to open ML models, in terms of enabling model publishers to sign their models during different stages of development and prove properties about the datasets they use.

Paperid:14

Authors:Aviv Ovadya · Kyle Redman · Oliver Smith · Luke Thorburn · Flynn Devine · Andrew Konya · Smitha Milli · Manon Revel · Quan Ze Chen · Kevin Feng · Amy Zhang · Bilva Chandra · Michiel Bakker · Atoosa Kasirzadeh

Title: Position: Democratic AI is Possible. The Democracy Levels Framework Shows How It Might Work.

Abstract:
This position paper argues that effectively “democratizing AI” requires democratic governance and alignment of AI, and that this is particularly valuable for decisions with systemic societal impacts. Initial steps — such as Meta’s Community Forums and Anthropic’s Collective Constitutional AI — have illustrated a promising direction, where democratic processes might be used to meaningfully improve public involvement and trust in critical decisions. To more concretely explore what increasingly democratic AI might look like, we provide a “Democracy Levels” framework and associated tools that: (i) define milestones toward meaningfully democratic, pluralistic, humancentered, and public-interest AI, (ii) can help guide organizations seeking to increase the legitimacy of their decisions on difficult AI governance and alignment questions, and (iii) support evaluation of such efforts.

Paperid:15

Authors:Yedi Zhang · Yufan Cai · Xinyue Zuo · Xiaokun Luan · Kailong Wang · Zhe Hou · Yifan Zhang · Zhiyuan Wei · Meng Sun · Jun Sun · Jing Sun · Jin Song Dong

Title: Position: Trustworthy AI Agents Require the Integration of Large Language Models and Formal Methods

Abstract:
Abstract:Large Language Models (LLMs) have emerged as a transformative AI paradigm, profoundly influencing daily life through their exceptional language understanding and contextual generation capabilities. Despite their remarkable performance, LLMs face a critical challenge: the propensity to produce unreliable outputs due to the inherent limitations of their learningbased nature. Formal methods (FMs), on the other hand, are a well-established computation paradigm that provides mathematically rigorous techniques for modeling, specifying, and verifying the correctness of systems. FMs have been extensively applied in mission-critical software engineering, embedded systems, and cybersecurity. However, the primary challenge impeding the deployment of FMs in real-world settings lies in their steep learning curves, the absence of user-friendly interfaces, and issues with efficiency and adaptability. This position paper outlines a roadmap for advancing the next generation of trustworthy AI systems by leveraging the mutual enhancement of LLMs and FMs. First, we illustrate how FMs, including reasoning and certification techniques, can help LLMs generate more reliable and formally certified outputs. Subsequently, we highlight how the advanced learning capabilities and adaptability of LLMs can significantly enhance the usability, efficiency, and scalability of existing FM tools. Finally, we show that unifying these two computation paradigms $-$ integrating the flexibility and intelligence of LLMs with the rigorous reasoning abilities of FMs $-$ has transformative potential for the development of trustworthy AI software systems. We acknowledge that this integration has the potential to enhance both the trustworthiness and efficiency of software engineering practices while fostering the development of intelligent FM tools capable of addressing complex yet real-world challenges.

Paperid:16

Authors:Kirsten Morehouse · Siddharth Swaroop · Weiwei Pan

Title: Position: Rethinking LLM Bias Probing Using Lessons from the Social Sciences

Abstract:
We argue that the new frontier of LLM bias probing can (and should) benefit from decades of social science research. While many LLM probes are direct applications of human ones, there remains a disconnect between psychological theory and LLM bias research. This introduces significant challenges: (1) we lack principled criteria for choosing appropriate probes, (2) we lack a system for reconciling conflicting results across probes, and (3) we lack formal frameworks for reasoning about when (and why) probe results will generalize to real user behavior. Existing taxonomies do not resolve these challenges. In this position paper, we describe actionable directions for systematizing LLM bias studies by connecting probes and the constructs they are intended to measure. We introduce EcoLevels – a framework grounded in social science for classifying and comparing bias probes. EcoLevels organizes probes along two features: the (abstraction) level at which bias is assessed and ecological validity (i.e. probetask generalization). We argue that EcoLevels can prevent end-users from drawing incorrect insights from LLM bias probing

Paperid:17

Authors:Sarah Hartman · Cheng Soon Ong · Julia Powles · Petra Kuhnert

Title: Position: We need responsible, application-driven (RAD) AI research

Abstract:
This position paper argues that achieving meaningful scientific and societal advances with artificial intelligence (AI) requires a responsible, applicationdriven approach (RAD) to AI research. As AI is increasingly integrated into society, AI researchers must engage with the specific contexts where AI is being applied. This includes being responsive to ethical and legal considerations, technical and societal constraints, and public discourse. We present the case for RAD-AI to drive research through a three-staged approach: (1) building transdisciplinary teams and people-centred studies; (2) addressing context-specific methods, ethical commitments, assumptions, and metrics; and (3) testing and sustaining efficacy through staged testbeds and a community of practice. We present a vision for the future of application-driven AI research to unlock new value through technically feasible methods that are adaptive to the contextual needs and values of the communities they ultimately serve.

Paperid:18

Authors:Yoonsoo Nam · Seok Hyeong Lee · Clémentine Dominé · Yeachan Park · Charles London · Wonyl Choi · Niclas Göring · Seungjai Lee

Title: Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)

Abstract:
In physics, complex systems are often simplified into minimal, solvable models that retain only the core principles. In machine learning, layerwise linear models (e.g., linear neural networks) act as simplified representations of neural network dynamics. These models follow the dynamical feedback principle, which describes how layers mutually govern and amplify each other's evolution. This principle extends beyond the simplified models, successfully explaining a wide range of dynamical phenomena in deep neural networks, including neural collapse, emergence, lazy and rich regimes, and grokking. In this position paper, we call for the use of layerwise linear models retaining the core principles of neural dynamical phenomena to accelerate the science of deep learning.

Paperid:19

Authors:Maya BechlerSpeicher · Ben Finkelshtein · Fabrizio Frasca · Luis Müller · Jan M Tönshoff · Antoine Siraudin · Viktor Zaverkin · Michael Bronstein · Mathias Niepert · Bryan Perozzi · Mikhail Galkin · Christopher Morris

Title: Position: Graph Learning Will Lose Relevance Due To Poor Benchmarks

Abstract:
While machine learning on graphs has demonstrated promise in drug design and molecular property prediction, significant benchmarking challenges hinder its further progress and relevance. Current benchmarking practices often lack focus on transformative, realworld applications, favoring narrow domains like two-dimensional molecular graphs over broader, impactful areas such as combinatorial optimization, databases, or chip design. Additionally, many benchmark datasets poorly represent the underlying data, leading to inadequate abstractions and misaligned use cases. Fragmented evaluations and an excessive focus on accuracy further exacerbate these issues, incentivizing overfitting rather than fostering generalizable insights. These limitations have prevented the development of truly useful graph foundation models. This position paper calls for a paradigm shift toward more meaningful benchmarks, rigorous evaluation protocols, and stronger collaboration with domain experts to drive impactful and reliable advances in graph learning research, unlocking the potential of graph learning.

Paperid:20

Authors:Giovanni Luca Marchetti · Vahid Shahverdi · Stefano Mereta · Matthew Trager · Kathlén Kohn

Title: Position: Algebra Unveils Deep Learning - An Invitation to Neuroalgebraic Geometry

Abstract:
In this position paper, we promote the study of function spaces parameterized by machine learning models through the lens of algebraic geometry. To this end, we focus on algebraic models, e.g. neural networks with polynomial activations, whose associated function spaces are semialgebraic varieties. We outline a dictionary between algebro-geometric invariants of these varieties, such as dimension, degree, and singularities, and fundamental aspects of machine learning, such as sample complexity, expressivity, training dynamics, and implicit bias. Along the way, we review the literature and discuss ideas beyond the algebraic domain. This work lays the foundations of a research direction bridging algebraic geometry and deep learning, that we refer to as neuroalgebraic geometry.

Paperid:21

Authors:Jan Kulveit · Raymond Douglas · Nora Ammann · Deger Turan · David Krueger · David Duvenaud

Title: Position: Humanity faces existential risk from gradual disempowerment

Abstract:
This paper examines the systemic risks posed by incremental advancements in artificial intelligence, developing the concept of `gradual disempowerment', in contrast to the abrupt takeover scenarios commonly discussed in AI safety. We analyze how even incremental improvements in AI capabilities can undermine human influence over largescale systems that society depends on, including the economy, culture, and nation-states. As AI increasingly replaces human labor and cognition in these domains, it can weaken both explicit human control mechanisms (like voting and consumer choice) and the implicit alignments with human preferences that often arise from societal systems' reliance on human participation to function. Furthermore, AI systems may amplify existing misalignments with human preferences by optimizing these systems more powerfully. These distortions across domains may be mutually reinforcing: economic power shapes cultural narratives and political decisions, while cultural shifts alter economic and political behavior. We argue that this dynamic could lead to an effectively irreversible loss of human influence over crucial societal systems, precipitating an existential catastrophe through the permanent disempowerment of humanity. This analysis suggests the need for both technical research and governance approaches that specifically address the risk of incremental erosion of human influence across interconnected societal systems.

Paperid:22

Authors:Jaeho Kim · Yunseok Lee · Seulki Lee

Title: Position: Towards a Sustainable AI Conference Peer Review System with Author Feedback and Reviewer Rewards

Abstract:
The peer review process in major artificial intelligence (AI) conferences faces unprecedented challenges with the surge of paper submissions (exceeding 10,000 submissions per venue), accompanied by growing concerns over review quality and reviewer responsibility. This position paper argues forthe need to transform the traditional oneway review system into a bi-directional feedback loop where authors evaluate review quality and reviewers earn formal accreditation, creating an accountability framework that promotes a sustainable, high-quality peer review system.The current review system can be viewed as an interaction between three parties: the authors, reviewers, and system (i.e., conference), where we posit that all three parties share responsibility for the current problems. However, issues with authors can only be addressed through policy enforcement and detection tools, and ethical concerns can only be corrected through self-reflection. As such, this paper focuses on reforming reviewer accountability with systematic rewards through two key mechanisms: (1) a two-stage bi-directional review system that allows authors to evaluate reviews while minimizing retaliatory behavior, (2) a systematic reviewer reward system that incentivizes quality reviewing. We ask the community's strong interest in these problems and the reforms that are needed to enhance the peer review process.

Paperid:23

Authors:Jaeho Kim · Yunseok Lee · Seulki Lee

Title: ICML Oral Position: Towards a Sustainable AI Conference Peer Review System with Author Feedback and Reviewer Rewards

Abstract:
The peer review process in major artificial intelligence (AI) conferences faces unprecedented challenges with the surge of paper submissions (exceeding 10,000 submissions per venue), accompanied by growing concerns over review quality and reviewer responsibility. This position paper argues forthe need to transform the traditional oneway review system into a bi-directional feedback loop where authors evaluate review quality and reviewers earn formal accreditation, creating an accountability framework that promotes a sustainable, high-quality peer review system.The current review system can be viewed as an interaction between three parties: the authors, reviewers, and system (i.e., conference), where we posit that all three parties share responsibility for the current problems. However, issues with authors can only be addressed through policy enforcement and detection tools, and ethical concerns can only be corrected through self-reflection. As such, this paper focuses on reforming reviewer accountability with systematic rewards through two key mechanisms: (1) a two-stage bi-directional review system that allows authors to evaluate reviews while minimizing retaliatory behavior, (2) a systematic reviewer reward system that incentivizes quality reviewing. We ask the community's strong interest in these problems and the reforms that are needed to enhance the peer review process.

Paperid:24

Authors:Marta An Kimmel · Mueed Rehman · Yonatan Bisk · Gary Fedder

Title: Position: You Can't CNC a NeRF

Abstract:
In this paper, we examine the manufacturability gap in stateof-the-art generative models for 3D object representations. Many models for generating 3D assets focused on rendering virtual content and do not consider the constraints of real-world manufacturing, such as milling, casting, or injection molding. We demonstrate existing generative models for computer-aided design representation do not generalize to datasets other than those used for training or to unmodified real, human-created objects. We identify current limitations with the current approaches, including missing manufacturing-readable semantics and ontology, the inability to decompose complex shapes into parameterized segments appropriate for computer-aided manufacturing, and a lack of appropriate scoring metrics to assess the generated output versus the true desired output from the dataset. The academic community could greatly impact real-world manufacturing by rallying around pathways to solve these challenges. We offer revised, more realistic datasets and baseline benchmarks as a step in targeting the challenge. In evaluating these datasets, we find that existing models are severely overfit to simpler data.

Paperid:25

Authors:Sayash Kapoor · Noam Kolt · Seth Lazar

Title: Position: Resist Platform-Controlled AI Agents and Champion User-Centric Agent Advocates

Abstract:
Language model agents could reshape how users navigate and act in digital environments. If controlled by platform companieswhich already dominate online search, communication, and commerce—platform agentswould intensify surveillance, exacerbate user lock-in, and further entrench the incumbent digital giants. This position paper argues that to resist the undesirable effects of platform agents, we should championagent advocates—agents that are controlled by users, serve the interests of users, and preserve user autonomy and choice. We identify key interventions to enable agent advocates: ensuring access to compute, developing interoperability protocols and safety standards, and implementing appropriate market regulations.

Paperid:26

Authors:Serena Booth

Title: Position: Strong Consumer Protection is an Inalienable Defense for AI Safety

Abstract:
Consumer protection laws are designed to protect consumers from unethical business practices. In this position paper, I argue that these laws serve an emergent dual purpose: if appropriately enforced and strengthened, consumer protection laws can serve as an inalienable defense for AI safety. These laws are well established and can be enforced and strengthened to incentivize businesses to design and deploy safer AI systems. This position runs counter to two prevailing trends in AI policy. The first alternative position is that AI safety requires an entirely new set of focused laws to protect humanity's prosperity. Though I find these efforts valuable, I argue that such focused laws are both hard to write and easy to skirt. The second alternative position is that consumer protection is nothing more than red tape; I argue that existing laws dating back many decades have already reigned in some nefarious business practices related to the development and deployment of AI, and that the litigious society of the United States is wellpositioned to use consumer protection laws to encourage new AI safety guardrails. This paper takes a tour of some existing consumer protection laws in the United States and their effects on the development and use of AI systems.

Paperid:27

Authors:Abeer Badawi · Md Tahmid Rahman Laskar · Jimmy Huang · Shaina Raza · Elham Dolatabadi

Title: Position: Beyond Assistance – Reimagining LLMs as Ethical and Adaptive Co-Creators in Mental Health Care

Abstract:
This position paper argues for a fundamental shift in how Large Language Models (LLMs) are integrated into the mental health care domain. We advocate for their role as cocreators rather than mere assistive tools. While LLMs have the potential to enhance accessibility, personalization, and crisis intervention, their adoption remains limited due to concerns about bias, evaluation, over-reliance, dehumanization, and regulatory uncertainties. To address these challenges, we propose two structured pathways: SAFE-I Implementation Guidelines for ethical and responsible deployment, and HAAS-E Evaluation Framework for multidimensional, human-centered assessment. SAFE-I provides a blueprint for data governance, adaptive model engineering, and real-world integration, ensuring LLMs align with clinical and ethical standards. HAAS-E introduces evaluation metrics that go beyond technical accuracy to measure trustworthiness, empathy, cultural sensitivity, and actionability. We call for the adoption of these structured approaches to establish a responsible and scalable model for LLM-driven mental health support, ensuring that AI complements—rather than replaces—human expertise.

Paperid:28

Authors:Yanzhou Pan · Jiayi Chen · Jiamin Chen · Zhaozhuo Xu · Denghui Zhang

Title: Position: Iterative Online-Offline Joint Optimization is Needed to Manage Complex LLM Copyright Risks

Abstract:
The infringement risks of LLMs have raised significant copyright concerns across different stages of the model lifecycle. While current methods often address these issues separately, this position paper argues that the LLM copyright challenges are inherently connected, and independent optimization of these solutions leads to theoretical bottlenecks. Building on this insight, we further argue that managing LLM copyright risks requires a systemic approach rather than fragmented solutions. In this paper, we analyze the limitations of existing methods in detail and introduce an iterative onlineoffline joint optimization framework to effectively manage complex LLM copyright risks. We demonstrate that this framework offers a scalable and practical solution to mitigate LLM infringement risks, and also outline new research directions that emerge from this perspective.

Paperid:29

Authors:Kevin Wei · Patricia Paskov · Sunishchal Dev · Michael Byun · Anka Reuel · Xavier Roberts-Gaal · Rachel Calcott · Evie Coxon · Chinmay Deshpande

Title: Position: Human Baselines in Model Evaluations Need Rigor and Transparency

Abstract:
This position paper argues that human baselines in foundation model evaluations must be more rigorous and more transparent to enable meaningful comparisons of human vs. AI performance.Human performance baselines are vital for the machine learning community, downstream users, and policymakers to interpret AI evaluations. Models are often claimed to achieve "superhuman" performance, but existing baselining methods are neither sufficiently rigorous nor sufficiently well-documented to robustly measure and assess performance differences. Based on a meta-review of the measurement theory and AI evaluation literatures, we derive a framework for assessing human baselining methods. We then use our framework to systematically review 113 human baselines in foundation model evaluations, identifying shortcomings in existing baselining methods. We hope our work can advance more rigorous AI evaluation practices that can better serve both the research community and policymakers. Data is available at: [GITHUB LINK].

Paperid:30

Authors:Jing Yang

Title: Position: The Artificial Intelligence and Machine Learning Community Should Adopt a More Transparent and Regulated Peer Review Process

Abstract:
The rapid growth of submissions to toptier Artificial Intelligence (AI) and Machine Learning (ML) conferences has prompted many venues to transition from closed to open review platforms. Some have fully embraced open peer reviews, allowing public visibility throughout the process, while others adopt hybrid approaches, such as releasing reviews only after final decisions or keeping reviews private despite using open peer review systems. In this work, we analyze the strengths and limitations of these models, highlighting the growing community interest in transparent peer review. To support this discussion, we examine insights from Paper Copilot, a website launched two years ago to aggregate and analyze AI / ML conference data while engaging a global audience. The site has attracted over 200,000 early-career researchers, particularly those aged 18–34 from 177 countries, many of whom are actively engaged in the peer review process. \textit{Drawing on our findings, this position paper advocates for a more transparent, open, and well-regulated peer review aiming to foster greater community involvement and propel advancements in the field.

Paperid:31

Authors:Sunayana Rane · Cyrus Kirkman · Graham Todd · Amanda Royka · Ryan Law · Erica Cartmill · Jacob Foster

Title: Position: Principles of Animal Cognition for LLM Evaluations: A Case Study on Transitive Inference

Abstract:
It has become increasingly challenging to understand and evaluate LLM capabilities as these models exhibit a broader range of behaviors. In this position paper, we argue that LLM researchers should draw on the lessons from another field which has developed a rich set of experimental paradigms and design practices for probing the behavior of complex intelligent systems: animal cognition. We present five core principles of evaluation drawn from animal cognition research, and explain how they provide invaluable guidance for understanding LLM capabilities and behavior. We ground these principles in an empirical case study, and show how they can already provide a richer picture of one particular reasoning capability: transitive inference.

Paperid:32

Authors:Sunayana Rane · Cyrus Kirkman · Graham Todd · Amanda Royka · Ryan Law · Erica Cartmill · Jacob Foster

Title: ICML Oral Position: Principles of Animal Cognition for LLM Evaluations: A Case Study on Transitive Inference

Abstract:
It has become increasingly challenging to understand and evaluate LLM capabilities as these models exhibit a broader range of behaviors. In this position paper, we argue that LLM researchers should draw on the lessons from another field which has developed a rich set of experimental paradigms and design practices for probing the behavior of complex intelligent systems: animal cognition. We present five core principles of evaluation drawn from animal cognition research, and explain how they provide invaluable guidance for understanding LLM capabilities and behavior. We ground these principles in an empirical case study, and show how they can already provide a richer picture of one particular reasoning capability: transitive inference.

Paperid:33

Authors:Ruth Elisabeth Appel

Title: Position: Generative AI Regulation Can Learn From Social Media Regulation

Abstract:
There is strong agreement that generative AI should be regulated, but strong disagreement on how to approach regulation. While some argue that AI regulation should mostly rely on extensions of existing laws, others argue that entirely new laws and regulations are needed to ensure that generative AI benefits society. In this position paper, we argue that the debates on generative AI regulation can be informed by the debates and evidence on social media regulation. For example, AI companies have faced allegations of political bias regarding the images and text their models produce, similar to the allegations social media companies have faced regarding content ranking on their platforms. First, we compare and contrast the affordances of generative AI and social media to highlight their similarities and differences. Then, we discuss specific policy recommendations based on the evolution of social media and their regulation. These recommendations include investments in (1) efforts to counter bias and perceptions thereof (e.g., via transparency, researcher access, oversight boards, democratic input, research studies), (2) specific areas of regulatory concern (e.g., youth wellbeing, election integrity) and trust and safety, (3) computational social science research, and (4) a more global perspective. Applying lessons learnt from social media regulation to generative AI regulation can save effort and time, and prevent avoidable mistakes.

Paperid:34

Authors:Ruth Elisabeth Appel

Title: ICML Oral Position: Generative AI Regulation Can Learn From Social Media Regulation

Abstract:
There is strong agreement that generative AI should be regulated, but strong disagreement on how to approach regulation. While some argue that AI regulation should mostly rely on extensions of existing laws, others argue that entirely new laws and regulations are needed to ensure that generative AI benefits society. In this position paper, we argue that the debates on generative AI regulation can be informed by the debates and evidence on social media regulation. For example, AI companies have faced allegations of political bias regarding the images and text their models produce, similar to the allegations social media companies have faced regarding content ranking on their platforms. First, we compare and contrast the affordances of generative AI and social media to highlight their similarities and differences. Then, we discuss specific policy recommendations based on the evolution of social media and their regulation. These recommendations include investments in (1) efforts to counter bias and perceptions thereof (e.g., via transparency, researcher access, oversight boards, democratic input, research studies), (2) specific areas of regulatory concern (e.g., youth wellbeing, election integrity) and trust and safety, (3) computational social science research, and (4) a more global perspective. Applying lessons learnt from social media regulation to generative AI regulation can save effort and time, and prevent avoidable mistakes.

Paperid:35

Authors:Oliver Eberle · Thomas McGee · Hamza Giaffar · Taylor Webb · Ida Momennejad

Title: Position: We Need An Algorithmic Understanding of Generative AI

Abstract:
What algorithms do LLMs actually learn and use to solve problems? Studies addressing this question are sparse, as research priorities are focused on improving performance through scale, leaving a theoretical and empirical gap in understanding emergent algorithms in LLMs. This position paper proposes AlgEval: a framework for systematic research into the algorithms that LLMs learn and use. AlgEval aims to uncover both algorithmic primitives, implemented via latent representations, attention weights, and inferencetime compute--and algorithmic composition--how these primitives are selected and combined to solve task-specific problems. We highlight potential methodological paths and a case study toward this goal, focusing on emergent search algorithms. Our case study illustrates both the formation of top-down hypotheses about candidate algorithms, and bottom-up tests of these hypotheses via circuit-level analysis of attention patterns and hidden states. The rigorous, systematic evaluation of how LLMs actually solve tasks provides an alternative to resource-intensive scaling, reorienting the field toward a principled understanding of underlying computations. This can in turn lead to more sample-efficient methods for training and improving performance, as well as novel architectures for end-to-end and multi-agent systems.

Paperid:36

Authors:Miriam Doh · Benedikt Höltgen · Piera Riccio · Nuria Oliver

Title: Position: The Categorization of Race in ML is a Flawed Premise

Abstract:
This position paper critiques the reliance on rigid racial taxonomies in machine learning, exposing their U.S.centric nature and lack of global applicability—particularly in Europe, where race categories are not commonly used. These classifications oversimplify racial identity, erasing the experiences of mixed-race individuals and reinforcing outdated essentialist views that contradict the social construction of race. We suggest research agendas in machine learning that move beyond categorical variables to better address discrimination and social inequality.

Paperid:37

Authors:Alan Jeffares · M van der Schaar

Title: Position: The Value of Deep Learning Phenomena Lies In Their Utility, Not Their Ubiquity

Abstract:
Developing a better understanding of surprising or counterintuitive phenomena has constituted a significant portion of deep learning research in recent years. These include double descent, grokking, and the lottery ticket hypothesis among many others. A natural motivation for this body of work appears to arise from the need to "resolve" or "understand" these phenomena on a narrow case-by-case basis. This position paper asserts that, in many prominent cases, there is little evidence to suggest that these phenomena appear in real-world applications and these efforts may be inefficient in driving progress in the broader field. In response, we advocate for a more intentionallypragmaticmethodology, prioritizing a focus on practical utility. We argue that they should be studied in a manner similar to natural phenomena (as in e.g. physics) where they can act as valuable challenges to our broader intuitions by taking ascientificapproach. This is in contrast to viewing them as individual challenges that require bespoke resolutions or explanations. This position is reinforced by analyzing the research outcomes of several prominent examples of these phenomena from the recent literature. Finally, we revisit the current norms in the research community in approaching these problems and propose practical recommendations for future research aiming to ensure that progress on deep learning phenomena is well aligned with the ultimate goal of progress in the broader field of deep learning.

Paperid:38

Authors:Alan Jeffares · M van der Schaar

Title: ICML Oral Position: The Value of Deep Learning Phenomena Lies In Their Utility, Not Their Ubiquity

Abstract:
Developing a better understanding of surprising or counterintuitive phenomena has constituted a significant portion of deep learning research in recent years. These include double descent, grokking, and the lottery ticket hypothesis among many others. A natural motivation for this body of work appears to arise from the need to "resolve" or "understand" these phenomena on a narrow case-by-case basis. This position paper asserts that, in many prominent cases, there is little evidence to suggest that these phenomena appear in real-world applications and these efforts may be inefficient in driving progress in the broader field. In response, we advocate for a more intentionallypragmaticmethodology, prioritizing a focus on practical utility. We argue that they should be studied in a manner similar to natural phenomena (as in e.g. physics) where they can act as valuable challenges to our broader intuitions by taking ascientificapproach. This is in contrast to viewing them as individual challenges that require bespoke resolutions or explanations. This position is reinforced by analyzing the research outcomes of several prominent examples of these phenomena from the recent literature. Finally, we revisit the current norms in the research community in approaching these problems and propose practical recommendations for future research aiming to ensure that progress on deep learning phenomena is well aligned with the ultimate goal of progress in the broader field of deep learning.

Paperid:39

Authors:Jacy Anthis · Ryan Liu · Sean Richardson · Austin Kozlowski · Bernard Koch · Erik Brynjolfsson · James Evans · Michael Bernstein

Title: Position: LLM Social Simulations Are a Promising Research Method

Abstract:
Accurate and verifiable large language model (LLM) simulations of human research subjects promise an accessible data source for understanding human behavior and training new AI systems. However, results to date have been limited, and few social scientists have adopted these methods. In this position paper, we argue that the promise of LLM social simulations can be achieved by addressing five tractable challenges. We ground our argument in a literature survey of empirical comparisons between LLMs and human research subjects, commentaries on the topic, and related work. We identify promising directions with prompting, finetuning, and complementary methods. We believe that LLM social simulations can already be used for exploratory research, such as pilot experiments for psychology, economics, sociology, and marketing. More widespread use may soon be possible with rapidly advancing LLM capabilities, and researchers should prioritize developing conceptual models and evaluations that can be iteratively deployed and refined at pace with ongoing AI advances.

Paperid:40

Authors:Min-Hsuan Yeh · Jeffrey Wang · Xuefeng Du · Seongheon Park · Leitian Tao · Shawn Im · Sharon Li

Title: Position: Challenges and Future Directions of Data-Centric AI Alignment

Abstract:
As AI systems become increasingly capable and influential, ensuring their alignment with human values, preferences, and goals has become a critical research focus. Current alignment methods primarily focus on designing algorithms and loss functions but often underestimate the crucial role of data. This paper advocates for a shift towards datacentric AI alignment, emphasizing the need to enhance the quality and representativeness of data used in aligning AI systems. In this position paper, we highlight key challenges associated with both human-based and AI-based feedback within the data-centric alignment framework. Through qualitative analysis, we identify multiple sources of unreliability in human feedback, as well as problems related to temporal drift, context dependence, and AI-based feedback failing to capture human values due to inherent model limitations. We propose future research directions, including improved feedback collection practices, robust data-cleaning methodologies, and rigorous feedback verification processes. We call for future research into these critical directions to ensure, addressing gaps that persist in understanding and improvingdata-centric alignment practices.

Paperid:41

Authors:Sebastian Bordt · Eric Raidl · Ulrike Luxburg

Title: Position: Rethinking Explainable Machine Learning as Applied Statistics

Abstract:
In the rapidly growing literature on explanation algorithms, it often remains unclear what precisely these algorithms are for and how they should be used. In this position paper, we argue for a novel and pragmatic perspective: Explainable machine learning needs to recognize its parallels with applied statistics. Concretely, explanations are statistics of highdimensional functions, and we should think about them analogously to traditional statistical quantities. Among others, this implies that we must think carefully about the matter of interpretation, or how the explanations relate to intuitive questions that humans have about the world. The fact that this is scarcely being discussed in research papers is one of the main drawbacks of the current literature. Luckily, the analogy between explainable machine learning and applied statistics provides a fruitful way for how research practices can be improved.

Paperid:42

Authors:John Hewitt · Robert Geirhos · Been Kim

Title: Position: We Can’t Understand AI Using our Existing Vocabulary

Abstract:
This position paper argues that, in order to understand AI, we cannot rely on our existing vocabulary of human words. Instead, we shouldstrive to develop neologisms: new words thatrepresent precise human concepts that we wantto teach machines, or machine concepts that weneed to learn. We start from the premise thathumans and machines have differing concepts.This means interpretability can be framed as acommunication problem: humans must be able toreference and control machine concepts, and communicate human concepts to machines. Creatinga shared humanmachine language through developing neologisms, we believe, could solve thiscommunication problem. Successful neologismsachieve a useful amount of abstraction: not toodetailed, so they’re reusable in many contexts, andnot too high-level, so they convey precise information. As a proof of concept, we demonstrate howa “length neologism” enables controlling LLMresponse length, while a “diversity neologism” allows sampling more variable responses. Takentogether, we argue that we cannot understand AIusing our existing vocabulary, and expanding itthrough neologisms creates opportunities for bothcontrolling and understanding machines better.

Paperid:43

Authors:Ahmed Alaa · Thomas Hartvigsen · Niloufar Golchini · Shiladitya Dutta · Frances Dean · Inioluwa Raji · Travis Zack

Title: Position: Medical Large Language Model Benchmarks Should Prioritize Construct Validity

Abstract:
Medical large language models (LLMs) research often makes bold claims, from encoding clinical knowledge to reasoning like a physician. These claims are usually backed by evaluation on competitive benchmarksa tradition inherited from mainstream machine learning. But how do we separate real progress from a leaderboard flex? Medical LLM benchmarks, much like those in other fields, are arbitrarily constructed using medical licensing exam questions. For these benchmarks to truly measure progress, they must accurately capture the real-world tasks they aim to represent. In this position paper, we argue that medical LLM benchmarks should—and indeed can—be empirically evaluated for their construct validity. In psychology, "construct validity" refers to how well a test measures the concept it is meant to assess. By drawing an analogy between LLM benchmarks and psychological tests, we explain how frameworks from this field can provide empirical foundations for validating benchmarks. To put this into practice, we use real-world clinical data in proof-of-concept experiments to evaluate popular medical LLM benchmarks and report gaps in their construct validity. Finally, we outline a vision for a new ecosystem of medical LLM evaluation centered around the creation of valid benchmarks.

Paperid:44

Authors:Ahmed Alaa · Thomas Hartvigsen · Niloufar Golchini · Shiladitya Dutta · Frances Dean · Inioluwa Raji · Travis Zack

Title: ICML Oral Position: Medical Large Language Model Benchmarks Should Prioritize Construct Validity

Abstract:
Medical large language models (LLMs) research often makes bold claims, from encoding clinical knowledge to reasoning like a physician. These claims are usually backed by evaluation on competitive benchmarksa tradition inherited from mainstream machine learning. But how do we separate real progress from a leaderboard flex? Medical LLM benchmarks, much like those in other fields, are arbitrarily constructed using medical licensing exam questions. For these benchmarks to truly measure progress, they must accurately capture the real-world tasks they aim to represent. In this position paper, we argue that medical LLM benchmarks should—and indeed can—be empirically evaluated for their construct validity. In psychology, "construct validity" refers to how well a test measures the concept it is meant to assess. By drawing an analogy between LLM benchmarks and psychological tests, we explain how frameworks from this field can provide empirical foundations for validating benchmarks. To put this into practice, we use real-world clinical data in proof-of-concept experiments to evaluate popular medical LLM benchmarks and report gaps in their construct validity. Finally, we outline a vision for a new ecosystem of medical LLM evaluation centered around the creation of valid benchmarks.

Paperid:45

Authors:Yan Shvartzshnaider · Vasisht Duddu

Title: Position: Contextual Integrity Washing for Language Models

Abstract:
Machine learning community is discovering Contextual Integrity (CI) as a useful framework to assess the privacy implications of large language models (LLMs). This is an encouraging development.The CI theory emphasizes sharing information in accordance withprivacy normsand can bridge the social, legal, political, and technical aspects essential for evaluating privacy in LLMs. However, this is also a good point to reflect on use of CI for LLMs.This position paper argues that existing literature adopts CI for LLMs without embracing the theory’s fundamental tenets, essentially amounting to a form of "CIwashing."CI-washing could lead to incorrect conclusions and flawed privacy-preserving designs. We clarify the four fundamental tenets of CI theory, systematize prior work on whether they deviate from these tenets, and highlight overlooked issues in experimental hygiene for LLMs (e.g., prompt sensitivity, positional bias).

Paperid:46

Authors:Sam Bowyer · Laurence Aitchison · Desi Ivanova

Title: Position: Don't use the CLT in LLM evals with fewer than a few hundred datapoints

Abstract:
Rigorous statistical evaluations of large language models (LLMs), including valid error bars and significance testing, are essential for meaningful and reliable performance assessment. Currently, when such statistical measures are reported, they typically rely on the Central Limit Theorem (CLT). In this position paper, we argue that while CLTbased methods for uncertainty quantification are appropriate when benchmarks consist of thousands of examples, they fail to provide adequate uncertainty estimates for LLM evaluations that rely on smaller, highly specialized benchmarks. In these small-data settings, we demonstrate that CLT-based methods perform very poorly, usually dramatically underestimating uncertainty (i.e. producing error bars that are too small). We give recommendations for alternative frequentist and Bayesian methods that are both easy to implement and more appropriate in these increasingly common scenarios.

Paperid:47

Authors:Ming Jin · Hyunin Lee

Title: Position: AI Safety Must Embrace an Antifragile Perspective

Abstract:
This position paper contends that modern AI research must adopt an antifragile perspective on safety—one in which the system's capacity to handle rare or outof-distribution (OOD) events adapts and expands over repeated exposures. Conventional static benchmarks and single-shot robustness tests overlook the reality that environments evolve and that models, if left unchallenged, can drift into maladaptation (e.g., reward hacking, over-optimization, or atrophy of broader capabilities). We argue that an antifragile approach---Rather than striving to rapidly reduce current uncertainties, the emphasis is on leveraging those uncertainties to better prepare for potentially greater, more unpredictable uncertainties in the future---is pivotal for the long-term reliability of open-ended ML systems. In this position paper, we first identify key limitations of static testing, including scenario diversity, reward hacking, and over-alignment. We then explore the potential of dynamic, antifragile solutions to manage rare events. Crucially, we advocate for a fundamental recalibration of the methods used to measure, benchmark, and continually improve AI safety over the long term by providing an ethical and practical guidelines to the AI safety community.

Paperid:48

Authors:Bruno Mlodozeniec · David Krueger · Richard E Turner

Title: Position: Probabilistic Modelling is Sufficient for Causal Inference

Abstract:
Causal inference is a key research area in machine learning, yet confusion reigns over the tools needed to tackle it. There are prevalent claims in the machine learning literature that you need a bespoke causal framework or notation to answer causal questions. In this paper, we make it clear that you can answer any causal inference question within the realm of probabilistic modelling and inference, without causalspecific tools or notation. Through concrete examples, we demonstrate how causal questions can be tackled by writing down the probability of everything. We argue for the advantages of the generality of the probabilistic modelling lens, when compared to bespoke causal frameworks. Lastly, we reinterpret causal tools as emerging from standard probabilistic modelling and inference, elucidating their necessity and utility.

Paperid:49

Authors:Bruno Mlodozeniec · David Krueger · Richard E Turner

Title: ICML Oral Position: Probabilistic Modelling is Sufficient for Causal Inference

Abstract:
Causal inference is a key research area in machine learning, yet confusion reigns over the tools needed to tackle it. There are prevalent claims in the machine learning literature that you need a bespoke causal framework or notation to answer causal questions. In this paper, we make it clear that you can answer any causal inference question within the realm of probabilistic modelling and inference, without causalspecific tools or notation. Through concrete examples, we demonstrate how causal questions can be tackled by writing down the probability of everything. We argue for the advantages of the generality of the probabilistic modelling lens, when compared to bespoke causal frameworks. Lastly, we reinterpret causal tools as emerging from standard probabilistic modelling and inference, elucidating their necessity and utility.

Paperid:50

Authors:Yucen Li · Daohan Lu · Polina Kirichenko · Shikai Qiu · Tim G. J. Rudner · C. Bayan Bruss · Andrew Wilson

Title: Position: Out-of-Distribution Detection Methods Answer the Wrong Questions

Abstract:
To detect distribution shifts and improve model safety, many outof-distribution (OOD) detection methods rely on the predictive uncertainty or features of supervised models trained on in-distribution data. In this position paper, we critically re-examine this popular family of OOD detection procedures, and we argue that these methods are fundamentally answering the wrong questions for OOD detection. There is no simple fix to this misalignment, since a classifier trained only on in-distribution classes cannot be expected to identify OOD points; for instance, a cat-dog classifier may confidently misclassify an airplane if it contains features that distinguish cats from dogs, despite generally appearing nothing alike. We find that uncertainty-based methods incorrectly conflate high uncertainty with being OOD, while feature-based methods incorrectly conflate far feature-space distance with being OOD. We show how these pathologies manifest as irreducible errors in OOD detection and identify common settings where these methods are ineffective. Additionally, interventions to improve OOD detection such as feature-logit hybrid methods, scaling of model and data size, epistemic uncertainty representation, and outlier exposure also fail to address this fundamental misalignment in objectives.

Paperid:51

Authors:Yunke Wang · Yanxi Li · Chang Xu

Title: Position: AI Scaling: From Up to Down and Out

Abstract:
AI Scaling has traditionally been synonymous with Scaling Up, which builds larger and more powerful models. However, the growing demand for efficiency, adaptability, and collaboration across diverse applications necessitates a broader perspective. This position paper presents a holistic framework for AI scaling, encompassing Scaling Up, Scaling Down, and Scaling Out. It argues that while Scaling Up of models faces inherent bottlenecks, the future trajectory of AI scaling lies in Scaling Down and Scaling Out. These paradigms address critical technical and societal challenges, such as reducing carbon footprint, ensuring equitable access, and enhancing crossdomain collaboration. We explore transformative applications in healthcare, smart manufacturing, and content creation, demonstrating how AI Scaling can enable breakthroughs in efficiency, personalization, and global connectivity. Additionally, we highlight key challenges, including balancing model complexity with interpretability, managing resource constraints, and fostering ethical development. By synthesizing these approaches, we propose a unified roadmap that redefines the future of AI research and application, paving the way for advancements toward Artificial General Intelligence (AGI).

Paperid:52

Authors:Sunayana Rane

Title: Position: AI's growing due process problem

Abstract:
AI systems are now ubiquitous in realworld decision-making. However, their use is often invisible and almost always difficult to understand for the ordinary people who now come into contact with AI regularly. As these AI-driven decision-making systems increasingly replace human counterparts, our ability to understand the reasons behind a decision, and to contest that decision fairly, is quickly being eroded. In the United States legal system due process includes the right to understand the reasons for certain major decisions and the right to openly contest those decisions. Everyone is entitled to due process under the law, and human decision-makers have been required to adhere to due process when making many important decisions that are now slowly being relegated to AI systems. Using two recent court decisions as a foundation, this paper takes the position that AI in its current form cannot guarantee due process, and therefore cannot and (should not) be used to make decisions that should be subject to due process. The supporting legal analysis investigates how the current lack of technical answers about the interpretability and causality of AI decisions, coupled with extreme trade secret protections severely limiting any exercise of the small amount of technical knowledge we do have, serve as a fatal anti-due-process combination. Finally, we discuss why technical researchers are vital to informing the legal process and protecting due process protections.

Paperid:53

Authors:D. Sculley · William Cukierski · Phil Culliton · Sohier Dane · Maggie Demkin · Ryan Holbrook · Addison Howard · Paul Mooney · Walter Reade · Meg Risdal · Nate Keating

Title: Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI Evaluation

Abstract:
In this position paper, we observe that empirical evaluation in Generative AI is at a crisis point since traditional ML evaluation and benchmarking strategies are insufficient to meet the needs of evaluating modern GenAI models and systems. There are many reasons for this, including the fact that these models typically have nearly unbounded input and output spaces, typically do not have a well defined ground truth target, and typically exhibit strong feedback loops and prediction dependence based on context of previous model outputs. On top of these critical issues, we argue that the problems ofleakageandcontaminationare in fact the most important and difficult issues to address for GenAI evaluations. Interestingly, the field of AI Competitions has developed effective measures and practices to combat leakage for the purpose of counteracting cheating by bad actors within a competition setting. This makes AI Competitions an especially valuable (but underutilized) resource. Now is time for the field to view AI Competitions as the gold standard for empirical rigor in GenAI evaluation, and to harness and harvest their results with according value.

Paperid:54

Authors:D. Sculley · William Cukierski · Phil Culliton · Sohier Dane · Maggie Demkin · Ryan Holbrook · Addison Howard · Paul Mooney · Walter Reade · Meg Risdal · Nate Keating

Title: ICML Oral Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI Evaluation

Abstract:
In this position paper, we observe that empirical evaluation in Generative AI is at a crisis point since traditional ML evaluation and benchmarking strategies are insufficient to meet the needs of evaluating modern GenAI models and systems. There are many reasons for this, including the fact that these models typically have nearly unbounded input and output spaces, typically do not have a well defined ground truth target, and typically exhibit strong feedback loops and prediction dependence based on context of previous model outputs. On top of these critical issues, we argue that the problems ofleakageandcontaminationare in fact the most important and difficult issues to address for GenAI evaluations. Interestingly, the field of AI Competitions has developed effective measures and practices to combat leakage for the purpose of counteracting cheating by bad actors within a competition setting. This makes AI Competitions an especially valuable (but underutilized) resource. Now is time for the field to view AI Competitions as the gold standard for empirical rigor in GenAI evaluation, and to harness and harvest their results with according value.

Paperid:55

Authors:Alex Gu · Naman Jain · WenDing Li · Manish Shetty Molahalli · Kevin Ellis · Koushik Sen · Armando Solar-Lezama

Title: Position: Tasks, Challenges, and Paths Towards AI for Software Engineering

Abstract:
AI for software engineering has made remarkable progress, becoming a notable success within generative AI. Despite this, achieving fully automated software engineering is still a significant challenge, requiring research efforts across both academia and industry. In this position paper, our goal is threefold. First, we provide a taxonomy of measures and tasks to categorize work towards AI software engineering. Second, we outline the key bottlenecks permeating today's approaches. Finally, we highlight promising paths towards making progress on these bottlenecks to guide future research in this rapidly maturing field.

Paperid:56

Authors:Hanqi Yan · Linhai Zhang · Jiazheng Li · Zhenyi Shen · Yulan He

Title: Position: LLMs Need a Bayesian MetaReasoning Framework for More Robust and Generalizable Reasoning

Abstract:
Large language models (LLMs) excel in many reasoning tasks but continue to face significant challenges, such as lack of robustness in reasoning, struggling with crosstask generalization, and inefficiencies in scaling up reasoning capabilities. Current training paradigms, including next-token prediction and reinforcement learning from human feedback, often fall short in adaptability to diverse reasoning tasks. Existing approaches, such as prompt optimization and iterative output refinement, offer performance improvement, but can be inefficient and lack effective generalization. To overcome these limitations, this position paper argues for a transformative shift in how LLMs approach reasoning. Drawing inspiration from cognitive science, particularly meta-reasoning theories such as Dual-Process Theory and Metacognitive Reasoning, we propose a Bayesian meta-reasoning framework for LLMs. Our approach integrates self-awareness, monitoring, evaluation, regulation, and meta-reflection, to enhance LLMs' ability to refine reasoning strategies and generalize across tasks. We revisit existing LLM reasoning methods, identify key challenges, and suggest directions for future research.

Paperid:57

Authors:Seungwook Han · Jyothish Pari · Samuel Gershman · Pulkit Agrawal

Title: Position: General Reasoning Requires Search From the Get-go

Abstract:
Large Language Models (LLMs) have demonstrated impressive realworld utility, exemplifying artificial useful intelligence (AUI). However, their ability to reason adaptively and robustly -- hallmarks of artificial general intelligence (AGI) -- remains limited. While LLMs seemingly succeed in commonsense reasoning, programming, and mathematics, they struggle to generalize algorithmic understanding across novel syntaxes and contexts. Our experiments with algorithmic tasks in esoteric programming languages reveal that LLMs' reasoning is overfitted to the training data, highlighting limitations in adaptability and transferability.In this position paper, we argue that the core issue in current LLMs lies in the coupling of reasoning and knowledge. To transition from AUI to AGI, we propose disentangling knowledge and reasoning through three key directions: (1) integrating RL during pretraining, (2) using artificial environments to learn reasoning priors and to make search in language more efficient, and (3) developing hybrid architectures that decouple memory from reasoning while maintaining dynamic interaction.

Paperid:58

Authors:Paul Youssef · Zhixue Zhao · Daniel Braun · Jörg Schlötterer · Christin Seifert

Title: Position: Editing Large Language Models Poses Serious Safety Risks

Abstract:
Large Language Models (LLMs) contain large amounts of facts about the world. These facts can become outdated over time, which has led to the development of knowledge editing methods (KEs) that can change specific facts in LLMs with limited side effects. This position paper argues that editing LLMs poses serious safety risks that have been largely overlooked. First, we note the fact that KEs are widely available, computationally inexpensive, highly performant, and stealthy makes them an attractive tool for malicious actors. Second, we discuss malicious use cases of KEs, showing how KEs can be easily adapted for a variety of malicious purposes. Third, we highlight vulnerabilities in the AI ecosystem that allow unrestricted uploading and downloading of updated models without verification. Fourth, we argue that a lack of social and institutional awareness exacerbates this risk, and discuss the implications for different stakeholders. We call on the community to (i) research tamperresistant models and countermeasures against malicious model editing, and (ii) actively engage in securing the AI ecosystem.

Paperid:59

Authors:Audrey Poinsot · Panayiotis Panayiotou · Alessandro Leite · Nicolas CHESNEAU · Özgür Şimşek · Marc Schoenauer

Title: Position: Causal Machine Learning Requires Rigorous Synthetic Experiments for Broader Adoption

Abstract:
Causal machine learning has the potential to revolutionize decisionmaking by combining the predictive power of machine learning algorithms with the theory of causal inference. However, these methods remain underutilized by the broader machine learning community, as current empirical evaluations of these methods do not permit assessment of their reliability and robustness, undermining their practical utility. Specifically, one of the principal criticisms made by the community is the extensive use of synthetic experiments. On the contrary, we argue that synthetic experiments are essential and necessary to precisely assess and understand the capabilities of causal machine learning methods. To substantiate our position, we critically review the current evaluation practices, spotlight their shortcomings, and propose a set of principles for conducting rigorous empirical analyses with synthetic data. Adopting our proposed principles will enable comprehensive evaluations that build trust in causal machine learning methods, driving their broader adoption and impactful real-world use.

Paperid:60

Authors:Yan Zhuang · Qi Liu · Zachary Pardos · Patrick Kyllonen · Jiyun Zu · Zhenya Huang · Shijin Wang · Enhong Chen

Title: Position: AI Evaluation Should Learn from How We Test Humans

Abstract:
As AI systems continue to evolve, their rigorous evaluation becomes crucial for their development and deployment. Researchers have constructed various largescale benchmarks to determine their capabilities, typically against a gold-standard test set and report metrics averaged across all items. However, this static evaluation paradigm increasingly shows its limitations, including high evaluation costs, data contamination, and the impact of low-quality or erroneous items on evaluation reliability and efficiency. In this Position, drawing from human psychometrics, we discuss a paradigm shift from static evaluation methods to adaptive testing. This involves estimating the characteristics or value of each test item in the benchmark, and tailoring each model's evaluation instead of relying on a fixed test set. This paradigm provides robust ability estimation, uncovering the latent traits underlying a model’s observed scores. This position paper analyze the current possibilities, prospects, and reasons for adopting psychometrics in AI evaluation. We argue thatpsychometrics, a theory originating in the 20th century for human assessment, could be a powerful solution to the challenges in today's AI evaluations.

Paperid:61

Authors:Michael Kirchhof · Gjergji Kasneci · Enkelejda Kasneci

Title: Position: Uncertainty Quantification Needs Reassessment for Large Language Model Agents

Abstract:
Large language models (LLMs) and chatbot agents are known to provide wrong outputs at times, and it was recently found that this can never be fully prevented. Hence, uncertainty quantification plays a crucial role, aiming to quantify the level of ambiguity in either one overall number or two numbers for aleatoric and epistemic uncertainty. This position paper argues that this traditional dichotomy of uncertainties is too limited for the open and interactive setup that LLM agents operate in when communicating with a user, and that we need to research avenues that enrich uncertainties in this novel scenario. We review the literature and find that popular definitions of aleatoric and epistemic uncertainties directly contradict each other and lose their meaning in interactive LLM agent settings. Hence, we propose three novel research directions that focus on uncertainties in such humancomputer interactions: Underspecification uncertainties, for when users do not provide all information or define the exact task at the first go, interactive learning, to ask follow-up questions and reduce the uncertainty about the current context, and output uncertainties, to utilize the rich language and speech space to express uncertainties as more than mere numbers. We expect that these new ways of dealing with and communicating uncertainties will lead to LLM agent interactions that are more transparent, trustworthy, and intuitive.

Paperid:62

Authors:Andreas Haupt · Erik Brynjolfsson

Title: Position: AI Should Not Be An Imitation Game: Centaur Evaluations

Abstract:
Benchmarks and evaluations are central to machine learning methodology and direct research in the field. Current evaluations test systems in the absence of humans. This position paper argues that the machine learning community should usecentaur evaluations, in which humans and AI jointly solve tasks. Centaur Evaluations refocus machine learning development toward human augmentation instead of human replacement. They allow for direct evaluation of humancentered desiderata, such as interpretability and helpfulness, and can be more challenging and realistic than existing evaluations. By shifting the focus fromautomationtowardcollaborationbetween humans and AI, centaur evaluations can drive progress toward more effective and human-augmenting AI systems.

Paperid:63

Authors:Lionel Wong · Ayman Ali · Raymond M Xiong · Shannon Shen · Yoon Kim · Monica Agrawal

Title: Position: Retrieval-augmented systems are currently dangerous medical communicators

Abstract:
Patients have long sought health information online, and increasingly, they are turning to generative AI to answer their healthrelated queries. Given the high stakes of the medical domain, techniques like retrieval-augmented generation and citation grounding have been widely promoted as methods to reduce hallucinations and improve the accuracy of AI-generated responses and have been widely adopted into search engines. However, we argue that even when these methods produce literally accurate content drawn from source documents sans hallucinations, they can still be highly misleading. Patients may derive significantly different interpretations from AI-generated outputs than they would from reading the original source material, let alone consulting a knowledgeable clinician. Through a large-scale query analysis on topics including disputed diagnoses and procedure safety, we support our argument with quantitative and qualitative evidence of the suboptimal answers resulting from current systems. In particular, we highlight how these models tend to decontextualize facts, omit critical relevant sources, and reinforce patient misconceptions or biases. We propose a series of recommendations---such as the incorporation of communication pragmatics and enhanced comprehension of source documents---that could help mitigate these issues and extend beyond the medical domain.

Paperid:64

Authors:Elliot Meyerson · Xin Qiu

Title: Position: Scaling LLM Agents Requires Asymptotic Analysis with LLM Primitives

Abstract:
Decomposing hard problems into subproblems often makes them easier and more efficient to solve. With the high cost of running LLMs at scale, there is an increasing effort to decompose systems into sets of LLMbased agents, each of whom can be delegated sub-tasks. However, this decomposition (even when automated) is often intuitive, e.g., based on how a human might assign roles to members of a human team. How close are these role decompositions to optimal? This position paper argues that asymptotic analysis with LLM primitives is needed to reason about the efficiency of such problem decompositions, and that insights from such analysis will unlock opportunities for scaling such systems. By treating the LLM forward pass as the atomic unit of computational cost, one can separate out the (often opaque) inner workings of a particular LLM from the inherent efficiency of how a set of LLMs are orchestrated to solve hard problems. In other words, if we want to scale the deployment of LLMs to the limit, instead of anthropomorphizing LLMs, asymptotic analysis with LLM primitives should be used to reason about and develop more powerful decompositions of large problems into LLM agents.

Paperid:65

Authors:Nikhil Kandpal · Colin Raffel

Title: Position: The Most Expensive Part of an LLM *should* be its Training Data

Abstract:
Large Language Model (LLM) training is an increasingly expensive endeavor due to growing computational requirements, hardware demands, energy costs, and engineering labor. Apart from training costs, an oftoverlooked (and seldom paid) cost is the human labor required to write the trillions of words of text used to train state-of-the-art LLMs. In this position paper, we aim to assign a monetary value to this labor and make the case that the most expensive part of producing an LLMshouldbe the compensation provided to training data producers for their work. To support this position we study 64 LLMs released between 2016 and 2024, breaking down both the cost of training the models as well as the hypothetical cost of creating the training data. Our analysis indicates that even with an extremely conservative estimate of how much compensation should be provided for the human labor that went into creating the training data, the costs of these models' training datasets are 1-3 orders of magnitude larger than the costs to train the models themselves. In the face of the massive gap between the value of training data and the current lack of compensation for its creation, we highlight and discuss research directions that could enable fairer practices in the future.

Paperid:66

Authors:Yuhe Guo · Huayi Tang · Jiahong Ma · Hongteng Xu · Zhewei Wei

Title: Position: Spectral GNNs Rely Less on Graph Fourier Basis than Conceived

Abstract:
Abstract:Spectral graph learning builds upon two foundations: Graph Fourier basis as its theoretical cornerstone,with polynomial approximation to enable practical implementation. While this framework has led to numerous successful designs, we argue that its effectiveness might stem from mechanisms different from its theoretical foundations. In this paper, we identify two fundamental issues that challenge our current understanding: (1) The graph Fourier basis $\mathbf{U}$ (eigenvectors of the normalized graph Laplacian) faces too many questions to truly serve its intended role, particularly in preserving its semantic properties of Fourier analysis; (2) The limitations preventing expressive filters are not merely practical constraints, but fundamental barriers that naturally protect stability and generalization. Importantly, the two issues entangle with each other. The second obscured the first: the natural avoidance of complex filters has prevented us from fully confronting the questions about $\mathbf{U}$'s role as a Fourier basis. This observation leads to our position: the effectiveness of spectral GNNs relies less on Graph Fourier basis than originally conceived, or, in other words, **spectral GNNs might not be so spectral**. The position leads us to at least two potential research interests: to incorporate a more semantically meaningful graph dictionary except for $\mathbf{U}$, and to reexamine the theoretical role of the introduced polynomial techniques.

Paperid:67

Authors:Golnaz Mesbahi · Parham Mohammad Panahi · Olya Mastikhina · Steven Tang · Martha White · Adam White

Title: Position: Lifetime tuning is incompatible with continual reinforcement learning

Abstract:
In continual RL we want agents capable of neverending learning, and yet our evaluation methodologies do not reflect this. The standard practice in RL is to assume unfettered access to the deployment environment for the full lifetime of the agent. For example, agent designers select the best performing hyperparameters in Atari by testing each for 200 million frames and then reporting results on 200 million frames. In this position paper, we argue and demonstrate the pitfalls of this inappropriate empirical methodology: lifetime tuning. We provide empirical evidence to support our position by testing DQN and SAC across several of continuing and non-stationary environments with two main findings: (1) lifetime tuning does not allow us to identify algorithms that work well for continual learning---all algorithms equally succeed; (2) recently developed continual RL algorithms outperform standard non-continual algorithms when tuning is limited to a fraction of the agent's lifetime. The goal of this paper is to provide an explanation for why recent progress in continual RL has been mixed and motivate the development of empirical practices that better match the goals of continual RL.

Paperid:68

Authors:Andy Zhang · Kevin Klyman · Yifan Mai · Yoav Levine · Yian Zhang · Rishi Bommasani · Percy Liang

Title: Position: Language model developers should report train-test overlap

Abstract:
Language models are extensively evaluated, but correctly interpreting evaluation results requires knowledge of traintest overlap, which refers to the extent to which the language model is trained on the very data it is being tested on. The public currently lacks adequate information about train-test overlap: most models have no public train-test overlap statistics, and third parties cannot directly measure train-test overlap since they do not have access to the training data. To make this clear, we document the practices of 30 models, finding that just 9 models report train-test overlap: 4 models release training data under open-source licenses, enabling the community to directly measure train-test overlap, and 5 models publish their train-test overlap methodology and statistics. By engaging with language model developers, we provide novel information about train-test overlap for three additional models. Overall, this position paper argues that language model developers should publish train-test overlap statistics and/or training data whenever they report evaluation results on public test sets. We hope our work increases transparency into train-test overlap to increase the community-wide trust in model evaluations.

Paperid:69

Authors:Rashid Mushkani · Hugo Berard · Allison Cohen · Shin Koseki

Title: Position: The Right to AI

Abstract:
This position paper proposes a “Right to AI,” which asserts that individuals and communities should meaningfully participate in the development and governance of the AI systems that shape their lives. Motivated by the increasing deployment of AI in critical domains and inspired by Henri Lefebvre's concept of the “Right to the City,” we reconceptualize AI as a societal infrastructure, rather than merely a product of expert design. In this paper, we critically evaluate how generative agents, largescale data extraction, and diverse cultural values bring new complexities to AI oversight. The paper proposes that grassroots participatory methodologies can mitigate biased outcomes and enhance social responsiveness. It asserts that data is socially produced and should be managed and owned collectively. Drawing on Sherry Arnstein’s Ladder of Citizen Participation and analyzing nine case studies, the paper develops a four-tier model for the Right to AI that situates the current paradigm and envisions an aspirational future. It proposes recommendations for inclusive data ownership, transparent design processes, and stakeholder-driven oversight. We also discuss market-led and state-centric alternatives and argue that participatory approaches offer a better balance between technical efficiency and democratic legitimacy.

Paperid:70

Authors:Kaiyu Yang · Gabriel Poesia · Jingxuan He · Wenda Li · Kristin Lauter · Swarat Chaudhuri · Dawn Song

Title: Position: Formal Mathematical Reasoning—A New Frontier in AI

Abstract:
AI for Mathematics (AI4Math) is intellectually intriguing and is crucial for AIdriven system design and verification. Extensive efforts on AI4Math have mirrored techniques in NLP, in particular, training large language models on carefully curated math datasets in text form. As a complementary yet less explored avenue, formal mathematical reasoning is grounded in formal systems such as proof assistants, which can verify the correctness of reasoning and provide automatic feedback. This position paper advocates formal mathematical reasoning as an indispensable component in future AI for math, formal verification, and verifiable generation. We summarize existing progress, discuss open challenges, and envision critical milestones to measure future success.

Paperid:71

Authors:Jillian Fisher · Ruth Elisabeth Appel · Chan Young Park · Yujin Potter · Liwei Jiang · Taylor Sorensen · Shangbin Feng · Yulia Tsvetkov · Margaret Roberts · Jennifer Pan · Dawn Song · Yejin Choi

Title: Position: Political Neutrality in AI is Impossible— But Here’s How to Approximate It

Abstract:
AI systems often exhibit political bias, influencing users' opinions and decisionmaking. While political neutrality—defined as the absence of bias—is seen as an ideal solution for fairness and safety, this position paper argues that true political neutrality is neither feasible nor universally desirable due to its subjective nature and the biases inherent in AI training data, algorithms, and user interactions. However, inspired by Joseph Raz's philosophical insight that "neutrality [...] can be a matter of degree" (Raz, 1986), we argue that striving for some neutrality remains essential for promoting balanced AI interactions and mitigating user manipulation. Therefore, we use the term "approximation" of political neutrality to shift the focus from unattainable absolutes to achievable, practical proxies. We propose eight techniques for approximating neutrality across three levels of conceptualizing AI, examining their trade-offs and implementation strategies. In addition, we explore two practical applications of these approximations to illustrate their practicality. Finally, we assess our framework on current large language models (LLMs) at the output level, providing a demonstration of how it can be evaluated. This work seeks to advance nuanced discussions of political neutrality in AI and promote the development of responsible, aligned language models.

Paperid:72

Authors:Jillian Fisher · Ruth Elisabeth Appel · Chan Young Park · Yujin Potter · Liwei Jiang · Taylor Sorensen · Shangbin Feng · Yulia Tsvetkov · Margaret Roberts · Jennifer Pan · Dawn Song · Yejin Choi

Title: ICML Oral Position: Political Neutrality in AI is Impossible— But Here’s How to Approximate It

Abstract:
AI systems often exhibit political bias, influencing users' opinions and decisionmaking. While political neutrality—defined as the absence of bias—is seen as an ideal solution for fairness and safety, this position paper argues that true political neutrality is neither feasible nor universally desirable due to its subjective nature and the biases inherent in AI training data, algorithms, and user interactions. However, inspired by Joseph Raz's philosophical insight that "neutrality [...] can be a matter of degree" (Raz, 1986), we argue that striving for some neutrality remains essential for promoting balanced AI interactions and mitigating user manipulation. Therefore, we use the term "approximation" of political neutrality to shift the focus from unattainable absolutes to achievable, practical proxies. We propose eight techniques for approximating neutrality across three levels of conceptualizing AI, examining their trade-offs and implementation strategies. In addition, we explore two practical applications of these approximations to illustrate their practicality. Finally, we assess our framework on current large language models (LLMs) at the output level, providing a demonstration of how it can be evaluated. This work seeks to advance nuanced discussions of political neutrality in AI and promote the development of responsible, aligned language models.

Paperid:73

Authors:Andrew C. Cullen · Paul MONTAGUE · Sarah Erfani · Benjamin Rubinstein

Title: Position: Certified Robustness Does Not (Yet) Imply Model Security

Abstract:
While certified robustness is widely promoted as a solution to adversarial examples in machine learning systems, significant challenges remain before these techniques can be meaningfully deployed in realworld applications. We identify critical gaps in current research, including the paradox of detection without distinction, the lack of clear criteria for practitioners to evaluate certification schemes, and the potential security risks arising from users' expectations surrounding ``guaranteed" robustness claims. This position paper is a call to arms for the certification research community, proposing concrete steps to address these fundamental challenges and advance the field toward practical applicability.

Paperid:74

Authors:Andrew C. Cullen · Paul MONTAGUE · Sarah Erfani · Benjamin Rubinstein

Title: ICML Oral Position: Certified Robustness Does Not (Yet) Imply Model Security

Abstract:
While certified robustness is widely promoted as a solution to adversarial examples in machine learning systems, significant challenges remain before these techniques can be meaningfully deployed in realworld applications. We identify critical gaps in current research, including the paradox of detection without distinction, the lack of clear criteria for practitioners to evaluate certification schemes, and the potential security risks arising from users' expectations surrounding ``guaranteed" robustness claims. This position paper is a call to arms for the certification research community, proposing concrete steps to address these fundamental challenges and advance the field toward practical applicability.

Paperid:75

Authors:Indradyumna Roy · Saswat Meher · Eeshaan Jain · Soumen Chakrabarti · Abir De

Title: Position: Graph Matching Systems Deserve Better Benchmarks

Abstract:
Data sets used in recent work on graph similarity scoring and matching tasks suffer from significant limitations. Using Graph Edit Distance (GED) as a showcase, we highlight pervasive issues such as traintest leakage and poor generalization, which have misguided the community's understanding and assessment of the capabilities of a method or model.These limitations arise, in part, because preparing labeled data is computationally expensive for combinatorial graph problems.We establish some key properties of GED that enable scalable data augmentation for training, and adversarial test set generation.Together, our analysis, experiments and insights establish new, sound guidelines for designing and evaluating future neural networks, and suggest open challenges for future research.

Paperid:76

Authors:Judy Hanwen Shen

Title: Position: Societal Impacts Research Requires Usage-Based Benchmark Development for Creative Composition Tasks

Abstract:
Foundation models that are capable of automating cognitive tasks represent a pivotal technological shift, yet their societal implications remain unclear. These systems promise exciting advances, yet they also risk flooding our information ecosystem with formulaic, homogeneous, and potentially misleading synthetic content. Developing benchmarks grounded in real use cases where these risks are most significant is therefore critical. Through a thematic analysis using 2 million language model user prompts, we identify creative composition tasks as a prevalent usage category where users seek help with personal tasks that require everyday creativity. Our finegrained analysis identifies mismatches between current benchmarks and usage patterns among these tasks. Crucially, we argue that the same use cases that currently lack thorough evaluations can lead to negative downstream impacts. This position paper argues that benchmarks focused on creative composition tasks is a necessary step towards understanding the societal harms of AI-generated content. We call for greater transparency in usage patterns to inform the development of new benchmarks that can effectively measure both the progress and the impacts of models with creative capabilities.

Paperid:77

Authors:Patrik Reizinger · Randall Balestriero · David Klindt · Wieland Brendel

Title: Position: An Empirically Grounded Identifiability Theory Will Accelerate Self Supervised Learning Research

Abstract:
SelfSupervised Learning (SSL) powers many current AI systems. As research interest and investment grow, the SSL design space continues to expand. The Platonic view of SSL, following the Platonic Representation Hypothesis (PRH), suggests that despite different methods and engineering approaches, all representations converge to the same Platonic ideal. However, this phenomenon lacks precise theoretical explanation. By synthesizing evidence from Identifiability Theory (IT), we show that the PRH can emerge in SSL. However, current IT cannot explain SSL's empirical success. To bridge the gap between theory and practice, we propose expanding IT into what we term Singular Identifiability Theory (SITh), a broader theoretical framework encompassing the entire SSL pipeline. SITh would allow deeper insights into the implicit data assumptions in SSL and advance the field towards learning more interpretable and generalizable representations. We highlight three critical directions for future research: 1) training dynamics and convergence properties of SSL; 2) the impact of finite samples, batch size, and data diversity; and 3) the role of inductive biases in architecture, augmentations, initialization schemes, and optimizers.

Paperid:78

Authors:Ossi Räisä · Boris van Breugel · M van der Schaar

Title: Position: All Current Generative Fidelity and Diversity Metrics are Flawed

Abstract:
Any method's development and practical application is limited by our ability to measure its reliability. The popularity of generative modeling emphasizes the importance of good synthetic data metrics. Unfortunately, previous works have found many failure cases in current metrics, for example lack of outlier robustness and unclear lower and upper bounds. We propose a list of desiderata for synthetic data metrics, and a suite of sanity checks: carefully chosen simple experiments that aim to detect specific and known generative modeling failure modes. Based on these desiderata and the results of our checks, we arrive at our position: all current generative fidelity and diversity metrics are flawed. This significantly hinders practical use of synthetic data. Our aim is to convince the research community to spend more effort in developing metrics, instead of models. Additionally, through analyzing how current metrics fail, we provide practitioners with guidelines on how these metrics should (not) be used.

Paperid:79

Authors:Yash Goel · Ayan Sengupta · Tanmoy Chakraborty

Title: Position: Enough of Scaling LLMs! Lets Focus on Downscaling

Abstract:
We challenge the dominant focus on neural scaling laws and advocate for a paradigm shift toward downscaling in the development of large language models (LLMs). While scaling laws have provided critical insights into performance improvements through increasing model and dataset size, we emphasize the significant limitations of this approach, particularly in terms of computational inefficiency, environmental impact, and deployment constraints. To address these challenges, we propose a holistic framework for downscaling LLMs that seeks to maintain performance while drastically reducing resource demands. This paper outlines practical strategies for transitioning away from traditional scaling paradigms, advocating for a more sustainable, efficient, and accessible approach to LLM development.

Paperid:80

Authors:Sanchaita Hazra · Bodhisattwa Prasad Majumder · Tuhin Chakrabarty

Title: Position: AI Safety should prioritize the Future of Work

Abstract:
Current efforts in AI safety prioritize filtering harmful content, preventing manipulation of human behavior, and eliminating existential risks in cybersecurity or biosecurity. While pressing, this narrow focus overlooks critical humancentric considerations that shape the long-term trajectory of a society. In this position paper, we identify the risks of overlooking the impact of AI on the future of work and recommend comprehensive transition support towards the evolution of meaningful labor with human agency. Through the lens of economic theories, we highlight the intertemporal impacts of AI on human livelihood and the structural changes in labor markets that exacerbate income inequality. Additionally, the closed-source approach of major stakeholders in AI development resembles rent-seeking behavior through exploiting resources, breeding mediocrity in creative labor, and monopolizing innovation. To address this, we argue in favor of a robust international copyright anatomy supported by implementing collective licensing that ensures fair compensation mechanisms for using data to train AI models. We strongly recommend a pro-worker framework of global AI governance to enhance shared prosperity and economic justice while reducing technical debt.

Paperid:81

Authors:Sanchaita Hazra · Bodhisattwa Prasad Majumder · Tuhin Chakrabarty

Title: ICML Oral Position: AI Safety should prioritize the Future of Work

Abstract:
Current efforts in AI safety prioritize filtering harmful content, preventing manipulation of human behavior, and eliminating existential risks in cybersecurity or biosecurity. While pressing, this narrow focus overlooks critical humancentric considerations that shape the long-term trajectory of a society. In this position paper, we identify the risks of overlooking the impact of AI on the future of work and recommend comprehensive transition support towards the evolution of meaningful labor with human agency. Through the lens of economic theories, we highlight the intertemporal impacts of AI on human livelihood and the structural changes in labor markets that exacerbate income inequality. Additionally, the closed-source approach of major stakeholders in AI development resembles rent-seeking behavior through exploiting resources, breeding mediocrity in creative labor, and monopolizing innovation. To address this, we argue in favor of a robust international copyright anatomy supported by implementing collective licensing that ensures fair compensation mechanisms for using data to train AI models. We strongly recommend a pro-worker framework of global AI governance to enhance shared prosperity and economic justice while reducing technical debt.

Paperid:82

Authors:Matthew Riemer · Zahra Ashktorab · Djallel Bouneffouf · Payel Das · Miao Liu · Justin Weisz · Murray Campbell

Title: Position: Theory of Mind Benchmarks are Broken for Large Language Models

Abstract:
This position paper argues that the majority of theory of mind benchmarks are broken because of their inability to directly test how large language models (LLMs) adapt to new partners. This problem stems from the fact that theory of mind benchmarks for LLMs are overwhelmingly inspired by the methods used to test theory of mind in humans and fall victim to a fallacy of attributing humanlike qualities to AI agents. We expect that humans will engage in a consistent reasoning process across various questions about a situation, but this is known to not be the case for current LLMs. Most theory of mind benchmarks only measure what we callliteral theory of mind: the ability to predict the behavior of others. Measuring this kind of reasoning is very informative in testing the ability of agents with self-consistent reasoning. However, it is important to note the distinction between this and what we actually care about when this self-consistency cannot be taken for granted. We call thisfunctional theory of mind: the ability to adapt to agents in-context following a rational response to predictions about their behavior. We find that top performing open source LLMs may display strong capabilities in literal theory of mind, depending on how they are prompted, but seem to struggle with functional theory of mind -- even when partner policies are exceedingly simple. Simply put, strong literal theory of mind performance does not necessarily imply strong functional theory of mind performance. Achieving functional theory of mind, particularly over long interaction horizons with a partner, is a significant challenge deserving a prominent role in any meaningful LLM theory of mind evaluation.

Paperid:83

Authors:Matej Pičulin · Bernarda Petek · Irena Ograjenšek · Erik Štrumbelj

Title: Position: Explainable AI Cannot Advance Without Better User Studies

Abstract:
In this position paper, we argue that user studies are key to evaluating explainable AI and that the current state of user studies is detrimental to the advancement of the field. We support this argument with a review of general and explainable AIspecific challenges, as well as an analysis of 607 explainable AI papers featuring user studies. To address this issue, we call for better-designed user studies and greater appreciation of high-quality user studies in the AI community.

Paperid:84

Authors:Shayne Longpre · Kevin Klyman · Ruth Elisabeth Appel · Sayash Kapoor · Rishi Bommasani · Michelle Sahar · Sean McGregor · Avijit Ghosh · Borhane Blili-Hamelin · Nathan Butters · Alondra Nelson · Amit Elazari · Andrew Sellars · Casey Ellis · Dane Sherrets · Dawn Song · Harley Geiger · Ilona Cohen · Lauren McIlvenny · Madhulika Srikumar · Mark Jaycox · Markus Anderljung · Nadine Johnson · Nicholas Carlini · Nicolas Miailhe · Nik Marda · Peter Henderson · Rebecca Portnoff · Rebecca Weiss · Victoria Westerhoff · Yacine Jernite · Rumman Chowdhury · Percy Liang · Arvind Narayanan

Title: Position: In-House Evaluation Is Not Enough. Towards Robust Third-Party Evaluation and Flaw Disclosure for General-Purpose AI

Abstract:
The widespread deployment of generalpurpose AI (GPAI) systems introduces significant new risks. Yet the infrastructure, practices, and norms for reporting flaws in GPAI systems lag far behind more mature industries, like software security. Based on a collaboration between experts from software security, machine learning, law, and policy, we identify key gaps in the evaluation and reporting of GPAI system flaws. Our position paper argues for three interventions to ultimately improve system safety: (1) the use of standardized AI system flaw reporting, (2) the adoption of flaw disclosure programs with researcher protections, and (3) infrastructure to coordinate flaw disclosure across stakeholders. First, we propose standardized AI flaw reports, to ease the process of submitting, reproducing, and triaging identified system flaws. Second, we propose GPAI system providers should adopt broadly-scoped flaw disclosure programs, borrowing from bug bounty programs, with legal safe harbors to protect both researchers and themselves. Third, we advocate for new infrastructure to coordinate flaw reports across many stakeholders, who may be impacted by certain flaws that commonly work on many independent GPAI systems (e.g., certain jailbreaks). By promoting a robust safety reporting and coordination in the GPAI ecosystem, these changes, if adopted, could ultimately improve AI system safety, security, and accountability.

Paperid:85

Authors:David Danhofer · Davide D'Ascenzo · Rafael Dubach · Tomaso A Poggio

Title: Position: A Theory of Deep Learning Must Include Compositional Sparsity

Abstract:
Overparametrized Deep Neural Networks (DNNs) have demonstrated remarkable success in a wide variety of domains too highdimensional for classical learners and shallow networks subject to the curse of dimensionality. However, open questions remain about fundamental principles, that govern the learning dynamics of DNNs. In this position paper, we argue that it is the ability of DNNs to exploit the compositionally sparse structure of the target function, that drives their success. As such, DNNs can leverage the property that most functions can be composed from a small set of constituent functions, each of which relies only on a low-dimensional subset of all inputs. We show that this property is shared by all efficiently Turing-computable functions and therefore highly likely present in all current learning problems. While some highly promising theoretical insights on questions concerned with approximation and generalization exist in the setting of compositionally sparse functions, several important questions on the learnability and optimization of DNNs remain open. Completing the picture on the role of compositional sparsity in deep learning is essential to a comprehensive theory of artificial - and even general - intelligence.

Paperid:86

Authors:Tobin South · Samuele Marro · Thomas Hardjono · Robert Mahari · Cedric Whitney · Alan Chan · Alex Pentland

Title: Position: AI Agents Need Authenticated Delegation

Abstract:
The rapid deployment of autonomous AI agents creates urgent challenges in the areas of authorization, accountability, and access control in task delegation. This position paper argues that authenticated and auditable delegation of authority to AI agents is a critical component of mitigating practical risks and unlocking the value of agents. To support this argument, we examine how existing web authentication and authorization protocols, as well as natural language interfaces to common access control mechanisms, can be extended to enable secure authenticated delegation of authority to AI agents. By extending OAuth 2.0 and OpenID Connect with agentspecific credentials and using transparent translation of natural language permissions into robust scoping rules across diverse interaction modalities, we outline how authenticated delegation can be achieved to enable clear chains of accountability while maintaining compatibility with established authentication and web infrastructure for immediate compatibility. This work contributes to ensuring that agentic AI systems perform only appropriate actions. It argues for prioritizing delegation infrastructure as a key component of AI agent governance and provides a roadmap for achieving this.

Paperid:87

Authors:Tobin South · Samuele Marro · Thomas Hardjono · Robert Mahari · Cedric Whitney · Alan Chan · Alex Pentland

Title: ICML Oral Position: AI Agents Need Authenticated Delegation

Abstract:
The rapid deployment of autonomous AI agents creates urgent challenges in the areas of authorization, accountability, and access control in task delegation. This position paper argues that authenticated and auditable delegation of authority to AI agents is a critical component of mitigating practical risks and unlocking the value of agents. To support this argument, we examine how existing web authentication and authorization protocols, as well as natural language interfaces to common access control mechanisms, can be extended to enable secure authenticated delegation of authority to AI agents. By extending OAuth 2.0 and OpenID Connect with agentspecific credentials and using transparent translation of natural language permissions into robust scoping rules across diverse interaction modalities, we outline how authenticated delegation can be achieved to enable clear chains of accountability while maintaining compatibility with established authentication and web infrastructure for immediate compatibility. This work contributes to ensuring that agentic AI systems perform only appropriate actions. It argues for prioritizing delegation infrastructure as a key component of AI agent governance and provides a roadmap for achieving this.

Paperid:88

Authors:Frithjof Gressmann · Ashley Chen · Lily Xie · Nancy Amato · Lawrence Rauchwerger

Title: Position: It Is Time We Test Neural Computation In Vitro

Abstract:
Recent advances in bioengineering have enabled the creation of biological neural networks in vitro, significantly reducing the cost, ethical hurdles, and complexity of experimentation with genuine biological neural computation. In this position paper, we argue that this trend offers a unique and timely opportunity to put our understanding of neural computation to the test. By designing artificial neural networks that can interact and control living neural systems, it is becoming possible to validate computational models beyond simulation and gain empirical insights to help unlock more robust and energyefficient next-generation AI systems. We provide an overview of key technologies, challenges, and principles behind this development and describe strategies and opportunities for novel machine learning research in this emerging field. We also discuss implications and fundamental questions that could be answered as this technology advances, exemplifying the longer-term impact of increasingly sophisticated in vitro neural networks.

Paperid:89

Authors:Samuel Gabriel Müller · Arik Reuter · Noah Hollmann · David Rügamer · Frank Hutter

Title: Position: The Future of Bayesian Prediction Is Prior-Fitted

Abstract:
Training neural networks on randomly generated artificial datasets leads to Bayesian models that capture the prior defined by the datasetgenerating distribution.Prior-data Fitted Networks (PFNs) are a class of methods designed to leverage this insight.In an era of rapidly increasing pre-training computational resources and a near stagnation in the generation of new real-world data in many applications, PFNs are poised to play a more important role across a wide range of applications.They enable the efficient allocation of pre-training compute to low-data scenarios.Originally applied to small Bayesian modeling tasks, the field of PFNs has significantly expanded to address more complex domains and larger datasets. This position paper argues that PFNs and other amortized inference approaches represent the future of Bayesian inference, leveraging amortized learning to tackle data-scarce problems. We thus believe they are a fruitful area of research. In this position paper, we explore their potential and directions to address their current limitations.

Paperid:90

Authors:Sebastin Santy · Prasanta Bhattacharya · Manoel Ribeiro · Kelsey Allen · Sewoong Oh

Title: Position: Sustaining human-generated data for ML requires shifting focus toward intrinsic human motivations

Abstract:
Progress in AI has relied on humangenerated data, from annotator marketplaces to the wider Internet. However, the widespread use of large language models now threatens the quality and integrity of human-generated data on these very platforms. We argue that this issue goes beyond the immediate challenge of filtering AI-generated content -- it reveals deeper flaws in how data collection systems are designed. Existing systems often prioritize speed, scale, and efficiency at the cost of intrinsic human motivation, leading to declining engagement and data quality. We propose that rethinking data collection systems to align with contributors' intrinsic motivations--rather than relying solely on external incentives--can help sustain high-quality data sourcing at scale while maintaining contributor trust and long-term participation.

Paperid:91

Authors:Tennison Liu · M van der Schaar

Title: Position: Self-Improving Agents Require Metacognitive Learning

Abstract:
Selfimproving agents aim to continuously acquire new capabilities with minimal supervision. However, current approaches face two key limitations: rigid self-improvement processes fail to generalize across domain/task distributions and increasing agent capabilities. We argue that self-improvement require metacognitive learning---an agent'sintrinsicability to actively evaluate, reflect, and adapt their learning processes dynamically. Drawing parallels to human metacognition, we introduce a formal framework with three components:metacognitive knowledge(self-assessment of capabilities, tasks, and strategies),metacognitive planning(deciding what and how to learn), andmetacognitive evaluation(reflecting on past experiences to refine learning). Analyzing current self-improving agents, we find they rely heavily on extrinsic metacognitive mechanisms, where fixed, human-designed self-improvement loops limit scalability and adaptability. Through analyzing each component, we contend that many of the ingredients to realize intrinsic metacognition are already present. Finally, we explore methods to optimally balance metacognition responsibilities between humans and agents, and identify the key technical hurdles that will enable self-improving agents to sustain efficient, generalizable, and aligned learning over time.

Paperid:92

Authors:Andrew Wilson

Title: Position: Deep Learning is Not So Mysterious or Different

Abstract:
Deep learning is often seen as different from other model classes by defying conventional notions of generalization. Popular examples of anomalous generalization behaviour include benign overfitting, double descent, and the success of overparametrization. This position paper argues that these phenomena are not distinct to neural networks, or particularly mysterious. Moreover, this generalization behaviour can be intuitively understood, and rigorously characterized using longstanding generalization frameworks such as PAC-Bayes and finite hypothesis bounds. We present soft inductive biases as a key unifying principle in explaining these phenomena: rather than restricting the hypothesis space to avoid overfitting, embrace a flexible hypothesis space, with a soft preference for simpler solutions that are consistent with the data. This principle can be encoded in many model classes, and thus deep learning is not as mysterious or different from other model classes as it might seem.

Paperid:93

Authors:Simone Drago · Marco Mussi · Alberto Maria Metelli

Title: Position: Constants are Critical in Regret Bounds for Reinforcement Learning

Abstract:
Mainstream research in theoretical RL is currently focused on designing online learning algorithms with regret bounds that match the corresponding regret lower bound up to multiplicative constants (and, sometimes, logarithmic terms). In this position paper, we constructively question this trend, arguing that algorithms should be designed to at least minimize the amount of unnecessary exploration, and we highlight the significant role constants play in algorithms' actual performances. This trend also exacerbates the misalignment between theoretical researchers and practitioners. As an emblematic example, we consider the case of regret minimization in finitehorizon tabular MDPs. Starting from the well-known UCBVI algorithm, we improve the bonus terms and the corresponding regret analysis. Additionally, we compare our version of UCBVI with both its original version and the state-of-the-art MVP algorithm. Our empirical validation successfully demonstrates how improving the multiplicative constants has significant positive effects on the actual empirical performances of the algorithm under analysis. This raises the question of whether ignoring constants when assessing whether algorithms match is the proper approach.

Paperid:94

Authors:Moming Duan · Mingzhe Du · Rui Zhao · Mengying Wang · Yinghui Wu · Nigel Shadbolt · Bingsheng He

Title: Position: Current Model Licensing Practices are Dragging Us into a Quagmire of Legal Noncompliance

Abstract:
The Machine Learning (ML) community has witnessed explosive growth, with millions of ML models being published on the Web. Reusing ML model components has been prevalent nowadays. Developers are often required to choose a license to publish and govern the use of their models. Popular options include Apache2.0, OpenRAIL (Responsible AI Licenses), Creative Commons Licenses (CCs), Llama2, and GPL-3.0. Currently, no standard or widely accepted best practices exist for model licensing. But does this lack of standardization lead to undesired consequences? Our answer is Yes. After reviewing the clauses of the most widely adopted licenses, we take the position thatcurrent model licensing practices are dragging us into a quagmire of legal noncompliance. To support this view, we explore the current practices in model licensing and highlight the differences between various model licenses. We then identify potential legal risks associated with these licenses and demonstrate these risks using examples from real-world repositories on Hugging Face. To foster a more standardized future for model licensing, we also propose a new draft of model licenses, MGLs, to address these challenges and promote better compliance.

Paperid:95

Authors:Moming Duan · Mingzhe Du · Rui Zhao · Mengying Wang · Yinghui Wu · Nigel Shadbolt · Bingsheng He

Title: ICML Oral Position: Current Model Licensing Practices are Dragging Us into a Quagmire of Legal Noncompliance

Abstract:
The Machine Learning (ML) community has witnessed explosive growth, with millions of ML models being published on the Web. Reusing ML model components has been prevalent nowadays. Developers are often required to choose a license to publish and govern the use of their models. Popular options include Apache2.0, OpenRAIL (Responsible AI Licenses), Creative Commons Licenses (CCs), Llama2, and GPL-3.0. Currently, no standard or widely accepted best practices exist for model licensing. But does this lack of standardization lead to undesired consequences? Our answer is Yes. After reviewing the clauses of the most widely adopted licenses, we take the position thatcurrent model licensing practices are dragging us into a quagmire of legal noncompliance. To support this view, we explore the current practices in model licensing and highlight the differences between various model licenses. We then identify potential legal risks associated with these licenses and demonstrate these risks using examples from real-world repositories on Hugging Face. To foster a more standardized future for model licensing, we also propose a new draft of model licenses, MGLs, to address these challenges and promote better compliance.

Paperid:96

Authors:Hanna Wallach · Meera Desai · A. Feder Cooper · Angelina Wang · Chad Atalla · Solon Barocas · Su Lin Blodgett · Alexandra Chouldechova · Emily Corvi · P. Alex Dow · Jean Garcia-Gathright · Alexandra Olteanu · Nicholas Pangakis · Stefanie Reed · Emily Sheng · Dan Vann · Jennifer Wortman Vaughan · Matthew Vogel · Hannah Washington · Abigail Z. Jacobs

Title: Position: Evaluating Generative AI Systems is a Social Science Measurement Challenge

Abstract:
The measurement tasks involved in evaluating generative AI (GenAI) systems are especially difficult, leading to what has been described as "a tangle of sloppy tests [and] applesto-oranges comparisons" (Roose, 2024). In this position paper, we argue that the ML community would benefit from learning from and drawing on the social sciences when developing and using measurement instruments for evaluating GenAI systems. Specifically, our position is that evaluating GenAI systems is a social science measurement challenge. We present a four-level framework, grounded in measurement theory from the social sciences, for measuring concepts related to the capabilities, behaviors, and impacts of GenAI. This framework has two important implications for designing and evaluating evaluations: First, it can broaden the expertise involved in evaluating GenAI systems by enabling stakeholders with different perspectives to participate in conceptual debates. Second, it brings rigor to both conceptual and operational debates by offering a set of lenses for interrogating the validity of measurement instruments and their resulting measurements.

Paperid:97

Authors:Borhane Blili-Hamelin · Christopher Graziul · Leif Hancox-Li · Hananel Hazan · El-Mahdi El-Mhamdi · Avijit Ghosh · Katherine Heller · Jacob Metcalf · Fabricio Murai · Eryk Salvaggio · ANDREW SMART · Todd Snider · Mariame Tighanimine · Talia Ringer · Margaret Mitchell · Shiri Dori-Hacohen

Title: Position: Stop treating `AGI' as the north-star goal of AI research

Abstract:
The AI research community plays a vital role in shaping the scientific, engineering, and societal goals of AI research. In this position paper, we argue that focusing on the highly contested topic of artificial general intelligence (`AGI') undermines our ability to choose effective goals. We identify six key trapsobstacles to productive goal setting---that are aggravated by AGI discourse: Illusion of Consensus, Supercharging Bad Science, Presuming Value-Neutrality, Goal Lottery, Generality Debt, and Normalized Exclusion. To avoid these traps, we argue that the AI research community needs to (1) prioritize specificity in engineering and societal goals, (2) center pluralism about multiple worthwhile approaches to multiple valuable goals, and (3) foster innovation through greater inclusion of disciplines and communities. Therefore, the AI research community needs to stop treating ``AGI'' as the north-star goal of AI research.

Paperid:98

Authors:Anish Dhir · Ruby Sedgwick · Avinash Kori · Ben Glocker · Mark van der Wilk

Title: Continuous Bayesian Model Selection for Multivariate Causal Discovery

Abstract:
Current causal discovery approaches require restrictive model assumptions in the absence of interventional data to ensure structure identifiability. These assumptions often do not hold in realworld applications leading to a loss of guarantees and poor performance in practice. Recent work has shown that, in the bivariate case, Bayesian model selection can greatly improve performance by exchanging restrictive modelling for more flexible assumptions, at the cost of a small probability of making an error. Our work shows that this approach is useful in the important multivariate case as well. We propose a scalable algorithm leveraging a continuous relaxation of the discrete model selection problem. Specifically, we employ the Causal Gaussian Process Conditional Density Estimator (CGP-CDE) as a Bayesian non-parametric model, using its hyperparameters to construct an adjacency matrix. This matrix is then optimised using the marginal likelihood and an acyclicity regulariser, giving the maximum a posteriori causal graph. We demonstrate the competitiveness of our approach, showing it is advantageous to perform multivariate causal discovery without infeasible assumptions using Bayesian model selection.

Paperid:99

Authors:Jingfeng Zhang · Prashanth Krishnamurthy · Naman Patel · Anthony Tzes · Farshad Khorrami

Title: MP-Nav: Enhancing Data Poisoning Attacks against Multimodal Learning

Abstract:
Despite the success of current multimodal learning at scale, its susceptibility to data poisoning attacks poses security concerns in critical applications. Attacker can manipulate model behavior by injecting maliciously crafted yet minute instances into the training set, stealthily mismatching distinct concepts. Recent studies have manifested the vulnerability by poisoning multimodal tasks such as TextImage Retrieval (TIR) and Visual Question Answering (VQA). However, the current attacking method only rely on random choice of concepts for misassociation and random instance selections for injecting the poisoning noise, which often achieves the suboptimal effect and even risks failure due to the dilution of poisons by the large number of benign instances. This study introduces MP-Nav (Multimodal Poison Navigator), a plug-and-play module designed to evaluate and even enhance data poisoning attacks against multimodal models. MP-Nav operates at both the concept and instance levels, identifying semantically similar concept pairs and selecting robust instances to maximize the attack efficacy. The experiments corroborate MP-Nav can significantly improve the efficacy of state-of-the-art data poisoning attacks such as AtoB and ShadowCast in multimodal tasks, and maintain model utility across diverse datasets. Notably, this study underscores the vulnerabilities of multimodal models and calls for the counterpart defenses.

Paperid:100

Authors:Wenju Sun · Qingyong Li · Yangliao Geng · Boyang Li

Title: CAT Merging: A Training-Free Approach for Resolving Conflicts in Model Merging

Abstract:
Multitask model merging offers a promising paradigm for integrating multiple expert models into a unified system without additional training. Existing state-of-the-art techniques, such as Task Arithmetic and its variants, merge models by accumulating task vectors—defined as the parameter differences between pre-trained and fine-tuned models. However, task vector accumulation is often hindered by knowledge conflicts, where conflicting components across different task vectors can lead to performance degradation during the merging process. To address this challenge, we proposeConflict-Aware Task Merging (CAT Merging), a novel training-free framework that selectively trims conflict-prone components from the task vectors. CAT Merging introduces several parameter-specific strategies, including projection for linear weights and masking for scaling and shifting parameters in normalization layers. Extensive experiments on vision and vision-language tasks demonstrate that CAT Merging effectively suppresses knowledge conflicts, achieving average accuracy improvements of up to 4.7% (ViT-B/32) and 2.0% (ViT-L/14) over state-of-the-art methods.

Paperid:101

Authors:Mohit Pandey · Gopeshh Subbaraj · Artem Cherkasov · Martin Ester · Emmanuel Bengio

Title: Pretraining Generative Flow Networks with Inexpensive Rewards for Molecular Graph Generation

Abstract:
Generative Flow Networks (GFlowNets) have recently emerged as a suitable framework for generating diverse and highquality molecular structures by learning from rewards treated as unnormalized distributions. Previous works in this framework often restrict exploration by using predefined molecular fragments as building blocks, limiting the chemical space that can be accessed. In this work, we introduce Atomic GFlowNets (A-GFNs), a foundational generative model leveraging individual atoms as building blocks to explore drug-like chemical space more comprehensively. We propose an unsupervised pre-training approach using drug-like molecule datasets, which teaches A-GFNs about inexpensive yet informative molecular descriptors such as drug-likeliness, topological polar surface area, and synthetic accessibility scores. These properties serve as proxy rewards, guiding A-GFNs towards regions of chemical space that exhibit desirable pharmacological properties. We further implement a goal-conditioned finetuning process, which adapts A-GFNs to optimize for specific target properties. In this work, we pretrain A-GFN on a subset of ZINC dataset, and by employing robust evaluation metrics we show the effectiveness of our approach when compared to other relevant baseline methods for a wide range of drug design tasks.

Paperid:102

Authors:Shihao Zou · Qingfeng Li · Wei Ji · Jingjing Li · Yongkui Yang · Guoqi Li · Chao Dong

Title: SpikeVideoFormer: An Efficient Spike-Driven Video Transformer with Hamming Attention and $\mathcal{O}(T)$ Complexity

Abstract:
Abstract:Spiking Neural Networks (SNNs) have shown competitive performance to Artificial Neural Networks (ANNs) in various vision tasks, while offering superior energy efficiency. However, existing SNNbased Transformers primarily focus on single-image tasks, emphasizing spatial features while not effectively leveraging SNNs' efficiency in video-based vision tasks. In this paper, we introduce SpikeVideoFormer, an efficient spike-driven video Transformer, featuring linear temporal complexity $\mathcal{O}(T)$. Specifically, we design a spike-driven Hamming attention (SDHA) which provides a theoretically guided adaptation from traditional real-valued attention to spike-driven attention. Building on SDHA, we further analyze various spike-driven space-time attention designs and identify an optimal scheme that delivers appealing performance for video tasks, while maintaining only linear temporal complexity. The generalization ability and efficiency of our model are demonstrated across diverse downstream video tasks, including classification, human pose tracking, and semantic segmentation. Empirical results show our method achieves state-of-the-art (SOTA) performance compared to existing SNN approaches, with over 15\% improvement on the latter two tasks. Additionally, it matches the performance of recent ANN-based methods while offering significant efficiency gains, achieving $\times 16$, $\times 10$ and $\times 5$ improvements on the three tasks, respectively. _Source code will be available._

Paperid:103

Authors:Wei Chen · Shigui Li · Jiacheng Li · Junmei Yang · John Paisley · Delu Zeng

Title: Dequantified Diffusion Schrödinger Bridge for Density Ratio Estimation

Abstract:
Abstract:Density ratio estimation is critical in applications involving fdivergences, yet existing methods struggle with multi-modal distributions or significant distributional discrepancies, leading to the \textit{density-chasm} problem. To address this, we propose Dequantified Diffusion-Bridge Interpolants (DDBI), which leverages diffusion processes to model smooth transitions between distributions. By integrating optimal transport theory, we extend DDBI to solve the Schrödinger-Bridge problem, resulting in Dequantified Schrödinger-Bridge Interpolants (DSBI).Combining these, we present the Dequantified Diffusion-bridge Density-Ratio Estimation ($\text{D}^3\text{RE}$) framework, which theoretically reduces estimation error in asymptotic density ratio estimation.Experiments demonstrate $\text{D}^3\text{RE}$'s in a variety of tasks, including mutual information estimation and density estimation.

Paperid:104

Authors:Jinda Lu · Junkang Wu · Jinghan Li · Xiaojun Jia · Shuo Wang · Yi-Fan Zhang · Junfeng Fang · Xiang Wang · Xiangnan He

Title: DAMO: Data- and Model-aware Alignment of Multi-modal LLMs

Abstract:
Direct Preference Optimization (DPO) has shown effectiveness in aligning multimodal large language models (MLLM) with human preferences. However, existing methods exhibit an imbalanced responsiveness to the data of varying hardness, tending to overfit on the easy-to-distinguish data while underfitting on the hard-to-distinguish data. In this paper, we propose Data- and Model-aware DPO (DAMO) to dynamically adjust the optimization process from two key aspects: (1) a data-aware strategy that incorporates data hardness, and (2) a model-aware strategy that integrates real-time model responses. By combining the two strategies, DAMO enables the model to effectively adapt to data with varying levels of hardness.Extensive experiments on five benchmarks demonstrate that DAMO not only significantly enhances the trustworthiness, but also improves the effectiveness over general tasks. For instance, on the Object HalBench, our DAMO-7B reduces response-level and mentioned-level hallucination by 90.0% and 95.3%, respectively, surpassing the performance of GPT-4V.

Paperid:105

Authors:Mingi Jung · Saehyung Lee · Eunji Kim · Sungroh Yoon

Title: Visual Attention Never Fades: Selective Progressive Attention ReCalibration for Detailed Image Captioning in Multimodal Large Language Models

Abstract:
Detailed image captioning is essential for tasks like data generation and aiding visually impaired individuals. Highquality captions require a balance between precision and recall, which remains challenging for current multimodal large language models (MLLMs). In this work, we hypothesize that this limitation stems from weakening and increasingly noisy visual attention as responses lengthen. To address this issue, we propose SPARC (Selective Progressive Attention ReCalibration), a training-free method that enhances the contribution of visual tokens during decoding. SPARC is founded on three key observations: (1) increasing the influence of all visual tokens reduces recall; thus, SPARC selectively amplifies visual tokens; (2) as captions lengthen, visual attention becomes noisier, so SPARC identifies critical visual tokens by leveraging attention differences across time steps; (3) as visual attention gradually weakens, SPARC reinforces it to preserve its influence. Our experiments, incorporating both automated and human evaluations, demonstrate that existing methods improve the precision of MLLMs at the cost of recall. In contrast, our proposed method enhances both precision and recall with minimal computational overhead.

Paperid:106

Authors:L. Elisa Celis · Lingxiao Huang · Nisheeth K. Vishnoi

Title: A Mathematical Framework for AI-Human Integration in Work

Abstract:
The rapid rise of Generative AI (GenAI) tools has sparked debates about their role in complementing or replacing human workers across job contexts. This paper approaches this question by presenting a mathematical framework for modeling jobs, workers, and metrics to evaluate workerjob fit. A key feature of our framework is its decomposition of skills into decision-level and action-level subskills, reflecting the distinct strengths of humans and AI. We analyze how changes in subskill abilities affect job success, identifying conditions for dramatic shifts in success probability. Furthermore, we establish sufficient conditions under which combining workers with complementary subskills significantly outperforms relying on a single worker. We demonstrate the framework’s practicality using data from O*NET and Big-Bench Lite, aligning real-world data with our framework through subskill-division methods. Our results highlight the role of GenAI tools in complementing human workers' skills rather than replacing them.

Paperid:107

Authors:Yixin Liu · Lie Lu · Jihui Jin · Lichao Sun · Andrea Fanelli

Title: XAttnMark: Learning Robust Audio Watermarking with Cross-Attention

Abstract:
The rapid proliferation of generative audio synthesis and editing technologies has raised significant concerns about copyright infringement, data provenance, and the spread of misinformation through deepfake audio. Watermarking offers a proactive solution by embedding imperceptible, identifiable, and traceable marks into audio content. While recent neural networkbased watermarking methods like WavMark and AudioSeal have improved robustness and quality, they struggle to achieve both robust detection and accurate attribution simultaneously. This paper introduces the Cross-Attention Robust Audio Watermark (XAttnMark), which bridges this gap by leveraging partial parameter sharing between the generator and the detector, a cross-attention mechanism for efficient message retrieval, and a temporal conditioning module for improved message distribution. Additionally, we propose a psychoacoustic-aligned temporal-frequency masking loss that captures fine-grained auditory masking effects, enhancing watermark imperceptibility. Our approach achieves state-of-the-art performance in both detection and attribution, demonstrating superior robustness against a wide range of audio transformations, including challenging generative editing with strong editing strength. This work represents a significant step forward in protecting intellectual property and ensuring the authenticity of audio content in the era of generative AI.

Paperid:108

Authors:Sifan Yang · Yuanyu Wan · Peijia Li · Yibo Wang · Xiao Zhang · Zhewei Wei · Lijun Zhang

Title: Dimension-Free Adaptive Subgradient Methods with Frequent Directions

Abstract:
Abstract:In this paper, we investigate the acceleration of adaptive subgradient methods through Frequent Directions (FD), a widelyused matrix sketching technique. The state-of-the-art regret bound exhibits a linear dependence on the dimensionality $d$, leading to unsatisfactory guarantees for high-dimensional problems. Additionally, it suffers from an $O(\tau^2 d)$ time complexity per round, which scales quadratically with the sketching size $\tau$. To overcome these issues, we first introduce an algorithm named FTSL, achieving a tighter regret bound that is independent of the dimensionality. The key idea is to integrate FD with adaptive subgradient methods under the primal-dual framework, and add the cumulative discarded information of FD back. To reduce its time complexity, we further utilize fast FD to expedite FTSL, yielding a better complexity of $O(\tau d)$ while maintaining the same regret bound. Moreover, to mitigate the computational cost for optimization problems involving matrix variables (e.g., training neural networks), we adapt FD to Shampoo, a popular optimization algorithm that accounts for the structure of decision, and give a novel analysis under the primal-dual framework. Our proposed method obtains an improved dimension-free regret bound alongside enhanced computational complexity. Experimental results have verified the efficiency and effectiveness of our approaches.

Paperid:109

Authors:Chendi Ge · Xin Wang · Zeyang Zhang · Hong Chen · Jiapei Fan · Longtao Huang · Hui Xue' · Wenwu Zhu

Title: D-MoLE: Dynamic Mixture of Curriculum LoRA Experts for Continual Multimodal Instruction Tuning

Abstract:
Continual multimodal instruction tuning (CMIT) is crucial for adapting Multimodal Large Language Models (MLLMs) to evolving tasks in realworld scenarios. However, most existing methods adopt a fixed architecture, struggling with adapting to new tasks due to static model capacity. In this paper, we propose evolving the architecture under parameter budgets for dynamic task adaptation in CMIT, a problem which remains unexplored in the literature. The problem presents two main challenges: 1) task architecture conflict, where different tasks require varying layer-wise adaptations, and 2) modality imbalance, where different tasks rely unevenly on modalities, leading to unbalanced updates. To address these challenges, we propose a novel Dynamic Mixture of Curriculum LoRA Experts (D-MoLE) method for CMIT, which automatically evolve MLLM's architecture with controlled parameter budgets to continually adapt to new tasks while retaining previously learned knowledge. Specifically, we propose a dynamic layer-wise expert allocator, which automatically allocates LoRA experts across layers to resolve architecture conflicts, and route instructions layer-wisely to facilitate knowledge sharing among experts. Then, we propose a gradient-based inter-modal continual curriculum, which adjusts the update ratio of each module in MLLM based on the difficulty of each modality within the task to alleviate the modality imbalance problem. Extensive experiments show that D-MoLE significantly outperforms state-of-the-art baselines, achieving a 15% average improvement over the best baseline. To the best of our knowledge, this is the first study of continual learning for MLLMs from an architectural perspective.

Paperid:110

Authors:Baijiong Lin · Weisen Jiang · Yuancheng Xu · Hao Chen · YINGCONG CHEN

Title: PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model

Abstract:
Multiobjective test-time alignment aims to adapt large language models (LLMs) to diverse multi-dimensional user preferences during inference without retraining. Recently, GenARM (Xu et al., 2025), a test-time alignment method, introduced the Autoregressive Reward Model (ARM) to effectively guide the generation of frozen LLMs during inference. When applied to multi-objective scenarios, GenARM needs multiple separately trained ARMs, each tailored to a specific preference dimension. GenARM faces two key limitations: the need for multiple ARMs increases the inference cost, and the separate training of ARMs leads to the misalignment between the guided generation and the use preferences.To address these issues, we propose Preference-aware ARM (PARM), a unified ARM conditioned on user preferences. PARM is implemented by our proposed Preference-Aware Bilinear Low-Rank Adaptation (PBLoRA), which employs a bilinear form to condition the ARM on user preferences, enabling it to dynamically manage trade-offs across multiple preference dimensions during inference. Experiments demonstrate that PARM reduces inference costs and improves alignment performance compared to GenARM. Additionally, PARM enables weak-to-strong guidance, allowing a smaller PARM to guide a larger frozen LLM without expensive training, offering a flexible solution for test-time preference adaptation.

Paperid:111

Authors:Antonio Almudévar · Jose Miguel Hernandez-Lobato · Sameer Khurana · Ricard Marxer · Alfonso Ortega

Title: Aligning Multimodal Representations through an Information Bottleneck

Abstract:
Contrastive losses have been extensively used as a tool for multimodal representation learning. However, it has been empirically observed that its use is not effective to learn an aligned representation space.In this paper, we argue that this phenomenon is caused by the presence of modalityspecific information in the representation space. Although some of the most widely used among the contrastive losses maximize the mutual information between representations of both modalities, they are not designed to remove the modality-specific information.We give a theoretical description of this problem through the lens of the Information Bottleneck Principle. We also empirically analyze how different hyperparameters affect to the emergence of this phenomenon in a controlled experimental setup.Finally, we propose a regularization term in the loss function that is derived by means of a variational approximation that aims to increase the representational alignment.We analyze in a set of controlled experiments and real-world applications the advantages of including this regularization term.

Paperid:112

Authors:Hongyi Liu · Rajarshi Saha · Zhen Jia · Youngsuk Park · Jiaji Huang · Shoham Sabach · Yu-Xiang Wang · George Karypis

Title: PROXSPARSE: REGULARIZED LEARNING OF SEMI-STRUCTURED SPARSITY MASKS FOR PRETRAINED LLMS

Abstract:
Large Language Models (LLMs) have demonstrated exceptional performance in natural language processing tasks, yet their massive size makes serving them inefficient and costly. Semistructured pruning has emerged as an effective method for model acceleration, but existing approaches are suboptimal because they focus on local, layer-wise optimizations using heuristic rules, failing to leverage global feedback. We present ProxSparse, a learning-based framework for mask selection enabled by regularized optimization. ProxSparse transforms the rigid, non-differentiable mask selection process into a smoother optimization procedure, allowing gradual mask exploration with flexibility. ProxSparse does not involve additional weight updates once the mask is determined. Our extensive evaluations on 7 widely used models show that ProxSparse consistently outperforms previously proposed semi-structured mask selection methods with significant improvement, demonstrating the effectiveness of our learned approach towards semi-structured pruning.

Paperid:113

Authors:Vincent P Grande · Michael Schaub

Title: Point-Level Topological Representation Learning on Point Clouds

Abstract:
Topological Data Analysis (TDA) allows us to extract powerful topological and higherorder information on the global shape of a data set or point cloud.Tools like Persistent Homology give a single complex description of the global structure of the point cloud.However, common machine learning applications like classification require point-level information and features.In this paper, we bridge this gap and propose a novel method to extract node-level topological features from complex point clouds using discrete variants of concepts from algebraic topology and differential geometry.We verify the effectiveness of these topological point features (TOPF) on both synthetic and real-world data and study their robustness under noise and heterogeneous sampling.

Paperid:114

Authors:Tejas Jayashankar · Jongha (Jon) Ryu · Gregory Wornell

Title: ScoreMix: One-Step Generative Model Training Made Simple via Score Estimation of Mixture Distributions

Abstract:
Abstract:We propose *ScoreMix*, a novel framework for training onestep generative models by minimizing a class of divergences called the$\alpha$-skew Jensen–Shannon divergence. At its core, ScoreMix estimates the score of mixture distributions between real and fake samples across multiple noise levels.Similar to consistency models, our approach supports both training from scratch and distillation using a pretrained diffusion model. It is simple to implement, requires minimal hyperparameter tuning, and ensures stable training. Experiments on CIFAR-10 and ImageNet 64×64 show that ScoreMix is competitive with and can even outperform existing methods.

Paperid:115

Authors:Liangchen Liu · Nannan Wang · Xi Yang · Xinbo Gao · Tongliang Liu

Title: Surrogate Prompt Learning: Towards Efficient and Diverse Prompt Learning for Vision-Language Models

Abstract:
Prompt learning is a cuttingedge parameter-efficient fine-tuning technique for pre-trained vision-language models (VLMs). Instead of learning a single text prompt, recent works have revealed that learning diverse text prompts can effectively boost the performances on downstream tasks, as the diverse prompted text features can comprehensively depict the visual concepts from different perspectives. However, diverse prompt learning demands enormous computational resources. This efficiency issue still remains unexplored. To achieve efficient and diverse prompt learning, this paper proposes a novel \textbf{Surrogate Prompt Learning (SurPL)} framework. Instead of learning diverse text prompts, SurPL directly generates the desired prompted text features via a lightweight \textbf{Surrogate Feature Generator (SFG)}, thereby avoiding the complex gradient computation procedure of conventional diverse prompt learning. Concretely, based on a basic prompted text feature, SFG can directly and efficiently generate diverse prompted features according to different pre-defined conditional signals. Extensive experiments indicate the effectiveness of the surrogate prompted text features, and show compelling performances and efficiency of SurPL on various benchmarks.

Paperid:116

Authors:Hao Gu · Wei Li · Lujun Li · Qiyuan Zhu · Mark Lee · Shengjie Sun · Wei Xue · Yike Guo

Title: Delta Decompression for MoE-based LLMs Compression

Abstract:
Abstract:Mixtureof-Experts (MoE) architectures in large language models (LLMs) achieve exceptional performance, but face prohibitive storage and memory requirements. To address these challenges, we present $D^2$-MoE, a new delta decompression compressor for reducing the parameters of MoE LLMs. Based on observations of expert diversity, we decompose their weights into a shared base weight and unique delta weights. Specifically, our method first merges each expert's weight into the base weight using the Fisher information matrix to capture shared components. Then, we compress delta weights through Singular Value Decomposition (SVD) by exploiting their low-rank properties.Finally, we introduce a semi-dynamical structured pruning strategy for the base weights, combining static and dynamic redundancy analysis to achieve further parameter reduction while maintaining input adaptivity. In this way, our $D^2$-MoE successfully compact MoE LLMs to high compression ratios without additional training. Extensive experiments highlight the superiority of our approach, with over 13\% performance gains than other compressors on Mixtral|Phi-3.5|DeepSeek|Qwen2 MoE LLMs at 40$\sim$60\% compression rates. Codes are available in supplementary materials.

Paperid:117

Authors:Ziru Wang · Mengmeng Wang · Jade Dai · Teli Ma · Guo-Jun Qi · Yong Liu · Guang Dai · Jingdong Wang

Title: DynaMind: Reasoning over Abstract Video Dynamics for Embodied Decision-Making

Abstract:
Integrating natural language instructions and visual perception with decisionmaking is a critical challenge for embodied agents. Existing methods often struggle to balance the conciseness of language commands with the richness of video content. To bridge the gap between modalities, we propose extracting key spatiotemporal patterns from video that capture visual saliency and temporal evolution, referred to as dynamic representation. Building on this, we introduce DynaMind, a framework that enhances decision-making through dynamic reasoning. Specifically, we design an adaptive FrameScorer to evaluate video frames based on semantic consistency and visual saliency, assigning each frame an importance score. These scores are used to filter redundant video content and synthesize compact dynamic representations. Leveraging these representations, we predict critical future dynamics and apply a dynamic-guided policy to generate coherent and context-aware actions. Extensive results demonstrate that DynaMind significantly outperforms the baselines across several simulation benchmarks and real-world scenarios.

Paperid:118

Authors:Ofri Eisen · Ron Dorfman · Kfir Levy

Title: Enhancing Parallelism in Decentralized Stochastic Convex Optimization

Abstract:
Decentralized learning has emerged as a powerful approach for handling large datasets across multiple machines in a communicationefficient manner. However, such methods often face scalability limitations, as increasing the number of machines beyond a certain point negatively impacts convergence rates. In this work, we proposeDecentralized Anytime SGD, a novel decentralized learning algorithm that significantly extends the critical parallelism threshold, enabling the effective use of more machines without compromising performance. Within the stochastic convex optimization (SCO) framework, we establish a theoretical upper bound on parallelism that surpasses the current state-of-the-art, allowing larger networks to achieve favorable statistical guarantees and closing the gap with centralized learning in highly connected topologies.

Paperid:119

Authors:Yangsibo Huang · Milad Nasr · Anastasios Angelopoulos · Nicholas Carlini · Wei-Lin Chiang · Christopher A. Choquette Choo · Daphne Ippolito · Matthew Jagielski · Katherine Lee · Ken Ziyu Liu · Ion Stoica · Florian Tramer · Chiyuan Zhang

Title: Exploring and Mitigating Adversarial Manipulation of Voting-Based Leaderboards

Abstract:
Abstract:It is now common to evaluate Large Language Models (LLMs) by having humans manually vote to evaluate model outputs, in contrast to typical benchmarks that evaluate knowledge or skill at some particular task. Chatbot Arena, the most popular benchmark of this type, ranks models by asking users to select the better response between two randomly selected models (without revealing which model was responsible for the generations). These platforms are widely trusted as a fair and accurate measure of LLM capabilities. In this paper, we show that if bot protection and other defenses are not implemented, these votingbased benchmarks are potentially vulnerable to adversarial manipulation. Specifically, we show that an attacker can alter the leaderboard (to promote their favorite model or demote competitors) at the cost of roughly a thousand votes (verified in a simulated, offline version of Chatbot Arena). Our attack consists of two steps: first, we show how an attacker can determine which model was used to generate a given reply with more than $95\%$ accuracy; and then, the attacker can use this information to consistently vote for (or against) a target model. Working with the Chatbot Arena developers, we identify, propose, and implement mitigations to improve the robustness of Chatbot Arena against adversarial manipulation, which, based on our analysis, substantially increases the cost of such attacks. Some of these defenses were present before our collaboration, such as bot protection with Cloudflare, malicious user detection, and rate limiting. Others, including reCAPTCHA and login are being integrated to strengthen the security in Chatbot Arena.

Paperid:120

Authors:Etowah Adams · Liam Bai · Minji Lee · Yiyang Yu · Mohammed AlQuraishi

Title: From Mechanistic Interpretability to Mechanistic Biology: Training, Evaluating, and Interpreting Sparse Autoencoders on Protein Language Models

Abstract:
Protein language models (pLMs) are powerful predictors of protein structure and function, learning through unsupervised training on millions of protein sequences. pLMs are thought to capture common motifs in protein sequences, but the specifics of pLM features are not well understood. Identifying these features would not only shed light on how pLMs work, but potentially uncover novel protein biology––studying the model to study the biology. Motivated by this, we train sparse autoencoders (SAEs) on the residual stream of a pLM, ESM2. By characterizing SAE features, we determine that pLMs use a combination of generic features and family-specific features to represent a protein. In addition, we demonstrate how known sequence determinants of properties such as thermostability and subcellular localization can be identified by linear probing of SAE features. For predictive features without known functional associations, we hypothesize their role in unknown mechanisms and provide visualization tools to aid their interpretation. Our study gives a better understanding of the limitations of pLMs, and demonstrates how SAE features can be used to help generate hypotheses for biological mechanisms.

Paperid:121

Authors:Zhuoling Li · Xiaogang Xu · Zhenhua Xu · Ser-Nam Lim · Hengshuang Zhao

Title: LARM: Large Auto-Regressive Model for Long-Horizon Embodied Intelligence

Abstract:
Recent embodied agents are primarily built based on reinforcement learning (RL) or large language models (LLMs). Among them, RL agents are efficient for deployment but only perform very few tasks. By contrast, giant LLM agents (often more than 1000B parameters) present strong generalization while demanding enormous computing resources. In this work, we combine their advantages while avoiding the drawbacks by conducting the proposed referee RL on our developed large autoregressive model (LARM). Specifically, LARM is built upon a lightweight LLM (fewer than 5B parameters) and directly outputs the next action to execute rather than text. We mathematically reveal that classic RL feedbacks vanish in long-horizon embodied exploration and introduce a giant LLM based referee to handle this reward vanishment during training LARM. In this way, LARM learns to complete diverse open-world tasks without human intervention. Especially, LARM successfully harvests enchanted diamond equipment in Minecraft, which demands significantly longer decision-making chains than the highest achievements of prior best methods.

Paperid:122

Authors:Yuxiao Wen

Title: Adversarial Combinatorial Semi-bandits with Graph Feedback

Abstract:
Abstract:In combinatorial semibandits, a learner repeatedly selects from a combinatorial decision set of arms, receives the realized sum of rewards, and observes the rewards of the individual selected arms as feedback. In this paper, we extend this framework to include \emph{graph feedback}, where the learner observes the rewards of all neighboring arms of the selected arms in a feedback graph $G$. We establish that the optimal regret over a time horizon $T$ scales as $\widetilde{\Theta}(S\sqrt{T}+\sqrt{\alpha ST})$, where $S$ is the size of the combinatorial decisions and $\alpha$ is the independence number of $G$. This result interpolates between the known regrets $\widetilde\Theta(S\sqrt{T})$ under full information (i.e., $G$ is complete) and $\widetilde\Theta(\sqrt{KST})$ under the semi-bandit feedback (i.e., $G$ has only self-loops), where $K$ is the number of actions. A technical ingredient is to realize a convexified action using a random decision vector with negative correlations, a result that may be of independent interest.

Paperid:123

Authors:Fan Wang · Feiyu Jiang · Zifeng Zhao · Yi Yu

Title: Transfer Learning for Nonparametric Contextual Dynamic Pricing

Abstract:
Dynamic pricing strategies are crucial for firms to maximize revenue by adjusting prices based on market conditions and customer characteristics. However, designing optimal pricing strategies becomes challenging when historical data are limited, as is often the case when launching new products or entering new markets. One promising approach to overcome this limitation is to leverage information from related products or markets to inform the focal pricing decisions. In this paper, we explore transfer learning for nonparametric contextual dynamic pricing under a covariate shift model, where the marginal distributions of covariates differ between source and target domains while the reward functions remain the same. We propose a novel Transfer Learning for Dynamic Pricing (TLDP) algorithm that can effectively leverage precollected data from a source domain to enhance pricing decisions in the target domain. The regret upper bound of TLDP is established under a simple Lipschitz condition on the reward function. To establish the optimality of TLDP, we further derive a matching minimax lower bound, which includes the target-only scenario as a special case and is presented for the first time in the literature. Extensive numerical experiments validate our approach, demonstrating its superiority over existing methods and highlighting its practical utility in real-world applications.

Paperid:124

Authors:Chongyu Fan · jinghan jia · Yihua Zhang · Anil Ramakrishna · Mingyi Hong · Sijia Liu

Title: Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond

Abstract:
The LLM unlearning technique has recently been introduced to comply with data regulations and address the safety and ethical concerns of LLMs by removing the undesired datamodel influence. However, state-of-the-art unlearning methods face a critical vulnerability: they are susceptible to ``relearning'' the removed information from a small number of forget data points, known as relearning attacks. In this paper, we systematically investigate how to make unlearned models robust against such attacks. For the first time, we establish a connection between robust unlearning and sharpness-aware minimization (SAM) through a unified robust optimization framework, in an analogy to adversarial training designed to defend against adversarial attacks. Our analysis for SAM reveals that smoothness optimization plays a pivotal role in mitigating relearning attacks. Thus, we further explore diverse smoothing strategies to enhance unlearning robustness. Extensive experiments on benchmark datasets, including WMDP and MUSE, demonstrate that SAM and other smoothness optimization approaches consistently improve the resistance of LLM unlearning to relearning attacks. Notably, smoothness-enhanced unlearning also helps defend against (input-level) jailbreaking attacks, broadening our proposal's impact in robustifying LLM unlearning.

Paperid:125

Authors:Tianyi Zhang · Junda Su · Aditya Desai · Oscar Wu · Zhaozhuo Xu · Anshumali Shrivastava

Title: Sketch to Adapt: Fine-Tunable Sketches for Efficient LLM Adaptation

Abstract:
Abstract:Adapting pretrained large language models (LLMs) is crucial but challenging due to their enormous size. Parameter-efficient fine-tuning (PEFT) techniques typically employ additive adapters applied to frozen model weights. To further reduce memory usage, model weights can be compressed through quantization. However, existing PEFT methods often yield suboptimal model quality due to restrictive assumptions, such as imposing low-rank constraints on adapters to reduce trainable parameters. We find that sketching, a popular data compression technique, can serve as an efficient adaptation strategy for LLMs while avoiding low-rank assumptions. We introduce SketchTune, a compressive adaptation strategy that compresses LLM weights into compact fine-tunable sketches, integrating compression and adaptation into a unified framework. This integration eliminates the need for complex two-path computation common in existing PEFT techniques, enabling faster and more memory-efficient training and inference. SketchTune is supported by mathematical insights into matrix classes that are better approximated using sketching rather than low-rank methods. Our rigorous evaluations with Llama-1/2/3 models demonstrate that SketchTune outperforms leading PEFT methods across diverse tasks including math problem-solving, common sense reasoning, and instruction following, while using substantially smaller base models and comparable trainable parameters. As a highlight, SketchTune outperforms LoRA, DoRA, and S2FT on commonsense and math benchmarks using 2.6-3.5$\times$ smaller base models and exceeds LoftQ in accuracy by 14.48\% on GSM8K with 7.3$\times$ fewer trainable parameters.

Paperid:126

Authors:Rujikorn Charakorn · Edoardo Cetin · Yujin Tang · Robert Lange

Title: Text-to-LoRA: Instant Transformer Adaption

Abstract:
While Foundation Models provide a general tool for rapid content creation, they regularly require taskspecific adaptation. Traditionally, this exercise involves careful curation of datasets and repeated fine-tuning of the underlying model. Fine-tuning techniques enable practitioners to adapt foundation models for many new applications but require expensive and lengthy training while being notably sensitive to hyper-parameter choices. To overcome these limitations, we introduce Text-to-LoRA (T2L), a model capable of adapting Large Language Models on the fly solely based on a natural language description of the target task. T2L is a hypernetwork trained to construct LoRAs in a single inexpensive forward pass. After training T2L on a suite of 9 pre-trained LoRA adapters (GSM8K, Arc, etc.), we show that the ad-hoc reconstructed LoRA instances match the performance of task-specific adapters across the corresponding test sets.Furthermore, T2L can compress hundreds of LoRA instances and zero-shot generalize to entirely unseen tasks. This approach provides a significant step towards democratizing the specialization of foundation models and enables language-based adaptation with minimal compute requirements. Our code and pre-trained checkpoints will be available upon publication.

Paperid:127

Authors:Kefan Dong · Tengyu Ma

Title: Beyond Limited Data: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving

Abstract:
A fundamental challenge in formal theorem proving by LLMs is the lack of highquality training data. Although reinforcement learning or expert iteration partially mitigates this issue by alternating between LLM generating proofs and finetuning them on correctly generated ones, performance quickly plateaus due to the scarcity of correct proofs (sparse rewards). To keep improving the models with limited data, we draw inspiration from mathematicians, who continuously develop new results, partly by proposing novel conjectures or exercises (which are often variants of known results) and attempting to solve them. We design the Self-play Theorem Prover (STP) that simultaneously takes on two roles, conjecturer and prover, each providing training signals to the other. The conjecturer is trained iteratively on previously generated conjectures that are barely provable by the current prover, which incentivizes it to generate increasingly challenging conjectures over time. The prover attempts to prove the conjectures with standard expert iteration. We evaluate STP with both Lean and Isabelle formal versifiers. With 19.8 billion tokens generated during the training in Lean, STP proves 26.3% of the statements in the LeanWorkbook dataset, doubling the previous best result of 13.2% achieved through expert iteration. The final model achieves state-of-the-art performance among whole-proof generation methods on miniF2F-test (61.1%, pass@3200), Proofnet-test (23.1%, pass@3200) and PutnamBench (8/644, pass@64).

Paperid:128

Authors:Zhehan Zhao · Lu Bai · Lixin Cui · Ming Li · Ziyu Lyu · Lixiang Xu · Yue Wang · Edwin Hancock

Title: ENAHPool: The Edge-Node Attention-based Hierarchical Pooling for Graph Neural Networks

Abstract:
Graph Neural Networks (GNNs) are powerful tools for graph learning, and one key operation is downsampling or pooling, which helps learn effective graph representations from node embeddings. In this paper, we propose a novel graph pooling operation, namely the EdgeNode Attention-based Hierarchical Pooling (ENAHPool), for GNNs to learn graph representations. Unlike previous cluster-based hierarchical pooling methods that rely on ambiguous node assignments and uniform aggregation of edge-node structural information, ENAHPool assigns each node entirely to a cluster. It then incorporates attention mechanisms to perform weighted aggregation, combining both node features within clusters and edge connectivity strengths between clusters at each pooling step. Additionally, since pooling operations are defined in conjunction with Message-Passing Neural Networks (MPNNs), we propose a Multi-Distance MPNN (MD-MPNN) architecture that enables nodes to directly receive feature information from neighbors at various distances, effectively mitigating the notorious over-squashing issue found in conventional MPNNs. Furthermore, it harnesses the valuable information implicit in edge connectivity strengths, further enhancing the performance of ENAHPool. Experiments demonstrate the effectiveness of the proposed ENAHPool operation associated with the MD-MPNN architecture.

Paperid:129

Authors:Ziyue Qiao · Qianyi Cai · Hao Dong · Jiawei Gu · Pengyang Wang · Meng Xiao · Xiao Luo · Hui Xiong

Title: GCAL: Adapting Graph Models to Evolving Domain Shifts

Abstract:
This paper addresses the challenge of graph domain adaptation on evolving, multiple outof-distribution (OOD) graphs.Conventional graph domain adaptation methods are confined to single-step adaptation, making them ineffective in handling continuous domain shifts and prone to catastrophic forgetting. This paper introduces the Graph Continual Adaptive Learning (GCAL) method, designed to enhance model sustainability and adaptability across various graph domains. GCAL employs a bilevel optimization strategy.The "adapt" phase uses an information maximization approach to fine-tune the model with new graph domains while re-adapting past memories to mitigate forgetting. Concurrently, the "generate memory", guided by a theoretical lower bound derived from information bottleneck theory, involves a variational memory graph generation module to condense original graphs into memories. Extensive experimental evaluations demonstrate that GCAL substantially outperforms existing methods in terms of adaptability and knowledge retention.

Paperid:130

Authors:Hongyao Chen · Tianyang Xu · Xiaojun Wu · Josef Kittler

Title: Hybrid Batch Normalisation: Resolving the Dilemma of Batch Normalisation in Federated Learning

Abstract:
Batch Normalisation (BN) is widely used in conventional deep neural network training to harmonise the inputoutput distributions for each batch of data.However, federated learning, a distributed learning paradigm, faces the challenge of dealing with non-independent and identically distributed data among the client nodes. Due to the lack of a coherent methodology for updating BN statistical parameters, standard BN degrades the federated learning performance.To this end, it is urgent to explore an alternative normalisation solution for federated learning. In this work, we resolve the dilemma of the BN layer in federated learning by developing a customised normalisation approach, Hybrid Batch Normalisation (HBN). HBN separates the update of statistical parameters (i.e. , means and variances used for evaluation) from that of learnable parameters (i.e. , parameters that require gradient updates), obtaining unbiased estimates of global statistical parameters in distributed scenarios. In contrast with the existing solutions, we emphasise the supportive power of global statistics for federated learning. The HBN layer introduces a learnable hybrid distribution factor, allowing each computing node to adaptively mix the statistical parameters of the current batch with the global statistics. Our HBN can serve as a powerful plugin to advance federated learning performance.It reflects promising merits across a wide range of federated learning settings, especially for small batch sizes and heterogeneous data.

Paperid:131

Authors:Jaehyun Kwak · Izaaz Inhar · Se-Young Yun · Sung-Ju Lee

Title: QuRe: Query-Relevant Retrieval through Hard Negative Sampling in Composed Image Retrieval

Abstract:
Composed Image Retrieval (CIR) retrieves relevant images based on a reference image and accompanying text describing desired modifications. However, existing CIR methods only focus on retrieving the target image and disregard the relevance of other images. This limitation arises because most methods employing contrastive learningwhich treats the target image as positive and all other images in the batch as negatives-can inadvertently include false negatives. This may result in retrieving irrelevant images, reducing user satisfaction even when the target image is retrieved. To address this issue, we propose Query-Relevant Retrieval through Hard Negative Sampling (QuRe), which optimizes a reward model objective to reduce false negatives. Additionally, we introduce a hard negative sampling strategy that selects images positioned between two steep drops in relevance scores following the target image, to effectively filter false negatives. In order to evaluate CIR models on their alignment with human satisfaction, we create Human-Preference FashionIQ (HP-FashionIQ), a new dataset that explicitly captures user preferences beyond target retrieval. Extensive experiments demonstrate that QuRe achieves state-of-the-art performance on FashionIQ and CIRR datasets while exhibiting the strongest alignment with human preferences on HP-FashionIQ dataset.

Paperid:132

Authors:Yi-Rui Yang · Chang-Wei Shi · Wu-Jun Li

Title: On the Tension between Byzantine Robustness and No-Attack Accuracy in Distributed Learning

Abstract:
Byzantinerobust distributed learning (BRDL), which refers to distributed learning with potential faulty workers (also known as Byzantine workers), has recently attracted much research attention. Robust aggregators are widely used in existing BRDL methods to obtain robustness against Byzantine workers. However, Byzantine workers do not always exist in applications. As far as we know, there is almost no existing work theoretically investigating the effect of using robust aggregators when there is no Byzantine worker. In view of this challenge, we theoretically analyze the aggregation error for robust aggregators when there is no Byzantine worker. Specifically, we show that the worst-case aggregation error without attack increases with the increase of the number of Byzantine workers that a robust aggregator can tolerate. The theoretical result reveals the tension between Byzantine robustness and no-attack accuracy. Furthermore, we provide lower bounds for the convergence rate of gradient descent with robust aggregators for non-convex objective functions and objective functions that satisfy the Polyak-Lojasiewicz (PL) condition, respectively. We also prove the tightness of the lower bounds. The lower bounds for convergence rate reveal similar tension between Byzantine robustness and no-attack accuracy. Empirical results further support our theoretical findings.

Paperid:133

Authors:TianChi Xing · Bonan Li · Congying Han · XINMIN QIU · Zicheng Zhang · Tiande Guo

Title: MIRROR: Make Your Multi-View Generation More Consistent with Training-Free Rectification

Abstract:
Multiview Diffusion has greatly advanced the development of 3D content creation by generating multiple images from distinct views, achieving remarkable photorealistic results. However, existing works are still vulnerable to inconsistent 3D geometric structures (commonly known as Janus Problem) and severe artifacts. In this paper, we introduce MIRROR, a versatile plug-and-play method that rectifies such inconsistencies in a training-free manner, enabling the acquisition of high-fidelity, realistic structures without compromising diversity. Our key idea focuses on tracing the motion trajectory of physical points across adjacent viewpoints, enabling rectifications based on neighboring observations of the same region. Technically, MIRROR comprises two core modules: Trajectory Tracking Module (TTM)for pixel-wise trajectory tracking that labels identical points across views, and Feature Rectification Module (FRM) for explicitly adjustment of each pixel embedding on noisy synthesized images by minimizing the distance to corresponding block features in neighboring views, thereby achieving consistent outputs. Extensive evaluations demonstrate that MIRROR can seamlessly integrate with a diverse range of off-the-shelf multi-view diffusion models, significantly enhancing both the consistency and the fidelity in an efficient way. Code will be released upon acceptance.

Paperid:134

Authors:Maxime Burchi · Radu Timofte

Title: Accurate and Efficient World Modeling with Masked Latent Transformers

Abstract:
Abstract:The Dreamer algorithm has recently obtained remarkable performance across diverse environment domains by training powerful agents with simulated trajectories. However, the compressed nature of its world model's latent space can result in the loss of crucial information, negatively affecting the agent's performance. Recent approaches, such as $\Delta$IRIS, address this limitation by training more accurate world models. However, these methods require training agents directly from pixels, which reduces training efficiency and prevents the agent from benefiting from the inner representations learned by the world model. In this work, we propose an alternative approach to world modeling that is both accurate and efficient. We introduce EMERALD (Efficient MaskEd latent tRAnsformer worLD model), a world model using a spatial latent state with MaskGIT predictions to generate accurate trajectories in latent space and improve the agent performance. On the Crafter benchmark, EMERALD achieves new state-of-the-art performance, becoming the first method to surpass human experts performance within 10M environment steps. Our method also succeeds to unlock all 22 Crafter achievements at least once during evaluation.

Paperid:135

Authors:Chenghua Liu · Minbo Gao · Zhengfeng Ji · Ying

Title: Quantum Speedup for Hypergraph Sparsification

Abstract:
Abstract:Graph sparsification serves as a foundation for many algorithms, such as approximation algorithms for graph cuts and Laplacian system solvers. As its natural generalization, hypergraph sparsification has recently gained increasing attention, with broad applications in graph machine learning and other areas. In this work, we propose the first quantum algorithm for hypergraph sparsification, addressing an open problem proposed by Apers and de Wolf (FOCS'20). For a weighted hypergraph with $n$ vertices, $m$ hyperedges, and rank $r$, our algorithm outputs a nearlinear size $\varepsilon$-spectral sparsifier in time $\widetilde O(r\sqrt{mn}/\varepsilon)$. This algorithm matches the quantum lower bound for constant $r$ and demonstrates quantum speedup when compared with the state-of-the-art $\widetilde O(mr)$-time classical algorithm. As applications, our algorithm implies quantum speedups for computing hypergraph cut sparsifiers, approximating hypergraph mincuts and hypergraph $s$-$t$ mincuts.

Paperid:136

Authors:Pratinav Seth · Michelle Lin · BREFO YAW · Jade Boutot · Mary Kang · David Rolnick

Title: Alberta Wells Dataset: Pinpointing Oil and Gas Wells from Satellite Imagery

Abstract:
Millions of abandoned oil and gas wells are scattered across the world, leaching methane into the atmosphere and toxic compounds into the groundwater. Many of these locations are unknown, preventing the wells from being plugged and their polluting effects averted. Remote sensing is a relatively unexplored tool for pinpointing abandoned wells at scale. We introduce the first largescale Benchmark dataset for this problem, leveraging high-resolution multi-spectral satellite imagery from Planet Labs. Our curated Dataset comprises over 213,000 wells (abandoned, suspended, and active) from Alberta, a region with especially high well density, sourced from the Alberta Energy Regulator and verified by domain experts. We evaluate baseline algorithms for well detection and segmentation, showing the promise of computer vision approaches but also significant room for improvement.

Paperid:137

Authors:Mengxiao Zhang · Yingfei Wang · Haipeng Luo

Title: Contextual Linear Bandits with Delay as Payoff

Abstract:
Abstract:A recent work by Schlisselberg et al. (2024) studies a delayas-payoff model for stochastic multi-armed bandits, where the payoff (either loss or reward) is delayed for a period that is proportional to the payoff. While this captures many real-world applications, the simple multi-armed bandit setting limits the practicality of their results. In this paper, we address this limitation by studying the delay-as-payoff model for contextual linear bandits. Specifically, we start from the case with a fixed action set and propose an efficient algorithm whose regret overhead compared to the standard no-delay case is only of order $D\Delta_{\max}\log T$, where $T$ is the total horizon, $D$ is the maximum delay, and $\Delta_{\max}$ is the maximum suboptimality gap. When payoff is loss, we also show further improvement of the bound, demonstrating a separation between reward and loss similar to Schlisselberg et al. (2024). Contrary to standard linear bandit algorithms that construct least squares estimator and confidence ellipsoid, the main novelty of our algorithm is to apply a phased arm elimination procedure by only picking the **volumetric spanners** of the action set, which addresses challenges arising from both payoff-dependent delays and large action sets. We further extend our results to the case with varying action sets by adopting the reduction from Hanna et al. (2023). Finally, we implement our algorithm and showcase its effectiveness and superior performance in experiments.

Paperid:138

Authors:Fabrizio Boninsegna · Jacob Imola · Anders Aamand · Abigail Gentle · Rasmus Pagh

Title: Lightweight Protocols for Distributed Private Quantile Estimation

Abstract:
Abstract:Distributed data analysis is a large and growing field driven by a massive proliferation of user devices, and by privacy concerns surrounding the centralised storage of data. We consider two \emph{adaptive} algorithms for estimating one quantile (e.g.~the median) when each user holds a single data point lying in a domain $[B]$ that can be queried once through a private mechanism; one under local differential privacy (LDP) and another for shuffle differential privacy (shuffleDP). In the adaptive setting we present an $\varepsilon$-LDP algorithm which can estimate any quantile within error $\alpha$ only requiring $O(\frac{\log B}{\varepsilon^2\alpha^2})$ users, and an $(\varepsilon,\delta)$-shuffle DP algorithm requiring only $\widetilde{O}((\frac{1}{\varepsilon^2}+\frac{1}{\alpha^2})\log B)$ users. Prior (nonadaptive) algorithms require more users by several logarithmic factors in $B$. We further provide a matching lower bound for adaptive protocols, showing that our LDP algorithm is optimal in the low-$\varepsilon$ regime. Additionally, we establish lower bounds against non-adaptive protocols which paired with our understanding of the adaptive case, proves a fundamental separation between these models.

Paperid:139

Authors:Daoyuan Chen · Haibin Wang · Yilun Huang · Ce Ge · Yaliang Li · Bolin Ding · Jingren Zhou

Title: Data-Juicer Sandbox: A Feedback-Driven Suite for Multimodal Data-Model Co-development

Abstract:
The emergence of multimodal large models has advanced artificial intelligence, introducing unprecedented levels of performance and functionality. However, optimizing these models remains challenging due to historically isolated paths of modelcentric and data-centric developments, leading to suboptimal outcomes and inefficient resource utilization. In response, we present a new sandbox suite tailored for integrated data-model co-development. This sandbox provides a feedback-driven experimental platform, enabling cost-effective iteration and guided refinement of both data and models. Our proposed ``Probe-Analyze-Refine'' workflow, validated through practical use cases on multimodal tasks such as image-text pre-training with CLIP, image-to-text generation with LLaVA-like models, and text-to-video generation with DiT-based models, yields transferable and notable performance boosts, such as topping the VBench leaderboard.Extensive experiments also uncover fruitful insights into the interplay between data quality, diversity, model behavior, and computational costs. All codes, datasets, and models are open-sourced to foster future research and applications that would otherwise be infeasible due to the lack of a dedicated co-development infrastructure.

Paperid:140

Authors:Mingzhe Yang · Sihao Lin · Changlin Li · Xiaojun Chang

Title: Let LLM Tell What to Prune and How Much to Prune

Abstract:
Large language models (LLMs) have revolutionized various AI applications. However, their billions of parameters pose significant challenges for practical deployment. Structured pruning is a hardwarefriendly compression technique and receives widespread attention. Nonetheless, existing literature typically targets a single structure of LLMs. We observe that different structure units within the LLM differently to inference cost and functionality. Therefore, pruning a single structure unit in isolation often results in an imbalance between the LLM's performance and efficiency after pruning. In addition, previous works mainly employ a prescribed pruning ratio. Since the significance of LLM modules may vary, it is ideal to distribute the pruning load to a specific structure unit according to its role within LLMs. To address the two issues, we propose a pruning method that targets multiple LLM modules with dynamic pruning ratios. Specifically, we find the intrinsic of LLM can guide us to determine the importance of each module and thus distribute the pruning load on demand, i.e., what to prune and how much to prune. This is achieved by quantifying the complex interactions within LLM. Through extensive experimentation on different language benchmarks and LLMs, we can demonstrate that our method effectively resolves the trade-off between efficiency and performance.

Paperid:141

Authors:Angelica Chen · Samuel Stanton · Frances Ding · Robert Alberstein · Andrew Watkins · Richard Bonneau · Vladimir Gligorijevic · Kyunghyun Cho · Nathan Frey

Title: Generalists vs. Specialists: Evaluating LLMs on Highly-Constrained Biophysical Sequence Optimization Tasks

Abstract:
Although large language models (LLMs) have shown promise in biomolecule optimization problems, they incur heavy computational costs and struggle to satisfy precise constraints. On the other hand, specialized solvers like LaMBO2 offer efficiency and fine-grained control but require more domain expertise. Comparing these approaches is challenging due to expensive laboratory validation and inadequate synthetic benchmarks. We address this by introducing Ehrlich functions, a synthetic test suite that captures the geometric structure of biophysical sequence optimization problems. With prompting alone, off-the-shelf LLMs struggle to optimize Ehrlich functions. In response, we propose \textsc{LLOME} (Language Model Optimization with Margin Expectation), a bilevel optimization routine for online black-box optimization.When combined with a novel preference learning loss, we find \textsc{LLOME} can not only learn to solve some Ehrlich functions, but can even outperform LaMBO-2 on moderately difficult Ehrlich variants. However, \textsc{LLOME} is comparable to \textsc{LaMBO-2} on very easy or difficult variants, exhibits some likelihood-reward miscalibration, and struggles without explicit rewards. Our results indicate LLMs can provide significant benefits in some cases, but specialized solvers are still competitive and incur less overhead.

Paperid:142

Authors:Xing Li · Zeyu Xing · Yiming Li · Linping Qu · Huiling Zhen · Yiwu Yao · Wulong Liu · Sinno Jialin Pan · Mingxuan Yuan

Title: KVTuner: Sensitivity-Aware Layer-wise Mixed Precision KV Cache Quantization for Efficient and Nearly Lossless LLM Inference

Abstract:
KV cache quantization can improve Large Language Models (LLMs) inference throughput and latency in long contexts and large batchsize scenarios while preserving LLMs effectiveness.However, current methods have three unsolved issues: overlooking layer-wise sensitivity to KV cache quantization, high overhead of online fine-grained decision-making, and low flexibility to different LLMs and constraints. Therefore, we thoroughly analyze the inherent correlation of layer-wise transformer attention patterns to KV cache quantization errors and study why key cache is more important than value cache for quantization error reduction. We further propose a simple yet effective framework KVTuner to adaptively search for the optimal hardware-friendly layer-wise KV quantization precision pairs for coarse-grained KV cache with multi-objective optimization and directly utilize the offline searched configurations during online inference.To reduce the computational cost of offline calibration, we utilize the intra-layer KV precision pair pruning and inter-layer clustering to reduce the search space. Experimental results show that we can achieve nearly lossless 3.25-bit mixed precision KV cache quantization for LLMs like Llama-3.1-8B-Instruct and 5.0-bit for sensitive models like Qwen2.5-7B-Instruct on mathematical reasoning tasks. The maximum inference throughput can be improved by 38.3% compared with KV8 quantization over various context lengths.

Paperid:143

Authors:Hailong Jiang · Jianfeng Zhu · Yao Wan · Bo Fang · Hongyu Zhang · Ruoming Jin · Qiang Guan

Title: Can Large Language Models Understand Intermediate Representations?

Abstract:
Intermediate Representations(IRs) are essential in compiler design and program analysis, yet their comprehension byLarge Language Models(LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.

Paperid:144

Authors:Haoyang Luo · Linwei Tao · Minjing Dong · Chang Xu

Title: Beyond One-Hot Labels: Semantic Mixing for Model Calibration

Abstract:
Model calibration seeks to ensure that models produce confidence scores that accurately reflect the true likelihood of their predictions being correct. However, existing calibration approaches are fundamentally tied to datasets of onehot labels implicitly assuming full certainty in all the annotations. Such datasets are effective for classification but provides insufficient knowledge of uncertainty for model calibration, necessitating the curation of datasets with numerically rich ground-truth confidence values. However, due to the scarcity of uncertain visual examples, such samples are not easily available as real datasets. In this paper, we introduce calibration-aware data augmentation to create synthetic datasets of diverse samples and their ground-truth uncertainty. Specifically, we present Calibration-aware Semantic Mixing (CSM), a novel framework that generates training samples with mixed class characteristics and annotates them with distinct confidence scores via diffusion models. Based on this framework, we propose calibrated reannotation to tackle the misalignment between the annotated confidence score and the mixing ratio during the diffusion reverse process. Besides, we explore the loss functions that better fit the new data representation paradigm. Experimental results demonstrate that CSM achieves superior calibration compared to the state-of-the-art calibration approaches.

Paperid:145

Authors:Kunal Jha · Wilka Carvalho · Yancheng Liang · Simon Du · Max Kleiman-Weiner · Natasha Jaques

Title: Cross-environment Cooperation Enables Zero-shot Multi-agent Coordination

Abstract:
Zeroshot coordination (ZSC), the ability to adapt to a new partner in a cooperative task, is a critical component of human-compatible AI. While prior work has focused on training agents to cooperate on a single task, these specialized models do not generalize to new tasks, even if they are highly similar. Here, we study how reinforcement learning on adistribution of environments with a single partnerenables learning general cooperative skills that support ZSC withmany new partners on many new problems. We introducetwoJax-based, procedural generators that create billions of solvable coordination challenges. We develop a new paradigm calledCross-Environment Cooperation (CEC), and show that it outperforms competitive baselines quantitatively and qualitatively when collaborating with real people. Our findings suggest that learning to collaborate across many unique scenarios encourages agents to develop general norms, which prove effective for collaboration with different partners. Together, our results suggest a new route toward designing generalist cooperative agents capable of interacting with humans without requiring human data.

Paperid:146

Authors:Hiroki Yanagisawa · Shunta Akiyama

Title: Survival analysis via density estimation

Abstract:
This paper introduces a novel framework for survival analysis by reinterpreting it as a form of density estimation. Our algorithm postprocesses density estimation outputs to derive survival functions, enabling the application of any density estimation model to effectively estimate survival functions. This approach broadens the toolkit for survival analysis and enhances the flexibility and applicability of existing techniques. Our framework is versatile enough to handle various survival analysis scenarios, including competing risk models for multiple event types. It can also address dependent censoring when prior knowledge of the dependency between event time and censoring time is available in the form of a copula. In the absence of such information, our framework can estimate the upper and lower bounds of survival functions, accounting for the associated uncertainty.

Paperid:147

Authors:Benjamin Ruben · William Tong · Hamza Chaudhry · Cengiz Pehlevan

Title: No Free Lunch from Random Feature Ensembles

Abstract:
Abstract:Given a fixed budget for total model size, one must choose between training a single large model or combining the predictions of multiple smaller models. We investigate this tradeoff for ensembles of random-feature ridge regression models in both the overparameterized and underparameterized regimes. Using asymptotically sharp risk estimates, we prove that when a fixed number of parameters is distributed among $K$ independently trained models, the ridge-optimized test risk increases with $K$.Consequently, a single large model achieves optimal performance. We then ask when *near*-optimal performance can be achieved by ensembles.In the overparameterized regime, we show that, to leading order, the test error depends on ensemble size and model size only through the total feature count, so that overparameterized ensembles consistently achieve near-optimal performance.To understand underparameterized ensembles, we derive scaling laws for the test risk as a function of total parameter count when the ensemble size and parameters per ensemble member are jointly scaled according to a ``growth exponent'' $\ell$. While optimal error scaling is always achieved by increasing model size with a fixed ensemble size, our analysis identifies conditions on the kernel and task eigenstructure under which near-optimal scaling laws can be obtained by joint scaling of ensemble size and model size.

Paperid:148

Authors:Samir Khaki · Xiuyu Li · Junxian Guo · Ligeng Zhu · Konstantinos N (Kostas) Plataniotis · Amir Yazdanbakhsh · Kurt Keutzer · Song Han · Zhijian Liu

Title: SparseLoRA: Accelerating LLM Fine-Tuning with Contextual Sparsity

Abstract:
Abstract:Finetuning LLMs is both computationally and memory-intensive. While parameter-efficient fine-tuning methods, such as DoRA, reduce the number of trainable parameters and lower memory usage, they do not decrease computational cost.In some cases, they even slow down fine-tuning.In this paper, we introduce *SparseLoRA*, a method that accelerates LLM fine-tuning through contextual sparsity.We propose a lightweight, training-free *SVD sparsity estimator* that dynamically selects a sparse subset of weights for loss and gradient computation.Also, we systematically analyze and address sensitivity across layers, tokens, and training steps.Our experimental results show that SparseLoRA reduces computational cost by up to **1.7$\times$** and a measured speedup of up to **1.4$\times$** while maintaining accuracy across various downstream tasks, including commonsense and arithmetic reasoning.We will release our code to encourage further research. This will support the development of fine-tuning methods that are both parameter- and computation-efficient.

Paperid:149

Authors:Qingqing Cao · Mahyar Najibi · Sachin Mehta

Title: CtrlSynth: Controllable Image Text Synthesis for Data-Efficient Multimodal Learning

Abstract:
Pretraining strong vision or multimodal foundation models like CLIP relies on largescale datasets (e.g., image-text pairs) that may be noisy, potentially misaligned, and have long-tail distributions. Previous work has shown promising results in augmenting datasets by generating synthetic samples. However, they only support domain-specific ad hoc use cases (like for image or text alone) and are limited in data diversity due to a lack of fine-grained control over the synthesis process. We design a controllable image-text synthesis pipeline called CtrlSynth to enable data-efficient multimodal learning and improve vision and multimodal models in various use cases. The key idea is to decompose the visual semantics of an image into basic elements, apply user-specified control policies (e.g. remove, add, replace operations), and recompose them to synthesize images or texts. The decompose and recompose feature in CtrlSynth allows users to control data synthesis in a fine-grained manner by defining customized control policies to manipulate the basic elements. CtrlSynth leverages the capabilities of pretrained foundation models such as large language models (LLMs) or diffusion models (DMs) to reason and recompose basic elements such that synthetic samples are natural and composed in diverse ways. CtrlSynth pipeline is training-free and has a modular design, making it easy to support different pretrained models. CtrlSynth pipeline is also closed-loop, meaning it can synthesize text data based on the image or vice versa. Our evaluation shows that CtrlSynth samples substantially improve zero-shot classification, image-text retrieval, and compositional reasoning performance of CLIP models. We will publicly release the code and pipeline for future research.

Paperid:150

Authors:Tianhao Wu · Janice Lan · Weizhe Yuan · Jiantao Jiao · JASON WESTON · Sainbayar Sukhbaatar

Title: Thinking LLMs: General Instruction Following with Thought Generation

Abstract:
LLMs are typically trained to answer user questions or follow instructions similarly to how human experts respond. However, in the standard alignment framework they lack the basic ability of explicit thinking before answering. Thinking is important for complex questions that require reasoning and planning but can be applied toanytask. We propose a training method for equipping existing LLMs with such thinking abilities for general instruction following without use of additional human data. We achieve this by an iterative search and optimization procedure that explores the space of possible thought generations, allowing the model to learn how to think without direct supervision. For each instruction, the thought candidates are scored using a judge model to evaluate their responses only, and then optimized via preference optimization. We show that this procedure leads to superior performance on AlpacaEval and Arena-Hard, and shows gains from thinking on non-reasoning categories such as marketing, health and general knowledge, in addition to more traditional reasoning & problem-solving tasks.

Paperid:151

Authors:Wa-Kin Lei · Jun-Cheng Chen · Shang-Tse Chen

Title: DRAG: Data Reconstruction Attack using Guided Diffusion

Abstract:
With the rise of large foundation models, split inference (SI) has emerged as a popular computational paradigm for deploying models across lightweight edge devices and cloud servers, addressing data privacy and computational cost concerns. However, most existing data reconstruction attacks have focused on smaller CNN classification models, leaving the privacy risks of foundation models in SI settings largely unexplored. To address this gap, we propose a novel data reconstruction attack based on guided diffusion, which leverages the rich prior knowledge embedded in a latent diffusion model (LDM) pretrained on a large-scale dataset. Our method performs iterative reconstruction on the LDM’s learned image prior, effectively generating high-fidelity images resembling the original data from their intermediate representations (IR). Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods, both qualitatively and quantitatively, in reconstructing data from deep-layer IRs of the vision foundation model. The results highlight the urgent need for more robust privacy protection mechanisms for large models in SI scenarios.

Paperid:152

Authors:Mingyu Kang · Yong Suk Choi

Title: Editable Noise Map Inversion: Encoding Target-image into Noise For High-Fidelity Image Manipulation

Abstract:
Textto-image diffusion models have achieved remarkable success in generating high-quality and diverse images. Building on these advancements, diffusion models have also demonstrated exceptional performance in text-guided image editing. A key strategy for effective image editing involves inverting the source image into editable noise maps associated with the target image. However, previous inversion methods face challenges in adhering closely to the target text prompt. The limitation arises because inverted noise maps, while enabling faithful reconstruction of the source image, restrict the flexibility needed for desired edits. To overcome this issue, we propose Editable Noise Map Inversion (ENM Inversion), a novel inversion technique that searches for optimal noise maps to ensure both content preservation and editability. We analyze the properties of noise maps for enhanced editability. Based on this analysis, our method introduces an editable noise refinement that aligns with the desired edits by minimizing the difference between the reconstructed and edited noise maps. Extensive experiments demonstrate that ENM Inversion outperforms existing approaches across a wide range of image editing tasks in both preservation and edit fidelity with target prompts. Our approach can also be easily applied to video editing, enabling temporal consistency and content manipulation across frames.

Paperid:153

Authors:Yiğit Narter · Alihan Hüyük · M van der Schaar · Cem Tekin

Title: Unified Screening for Multiple Diseases

Abstract:
Current screening programs that focus on improving patient health while minimizing screening costs are tailored for individual diseases. Designing unified screening programs for multiple diseases requires carefully balancing competing disease risks, which is an open problem. In this work, we address this problem by casting unified screening as a referral problem, in which we choose to activate a subset of screening policies for individual diseases by accounting for competing risks that influence patient outcomes. We introduce a novel optimization framework that incorporates disease risks, budget constraints, and diagnostic error limits and characterize the structural properties of the optimal referral policy. For the unified screening of two diseases, we show that the optimal activation threshold for the screening of one disease depends on the risk of the other, resulting in decision boundaries with distinct riskdependent profiles. We compare our unified model with independent screening programs that apply isolated activation thresholds for screening of each disease. Our approach optimizes screening decisions collectively, improving overall survival outcomes, particularly for patients with high disease risks.

Paperid:154

Authors:Julien Pourcel · Cédric Colas · Pierre-Yves Oudeyer

Title: Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI

Abstract:
Many program synthesis tasks prove too challenging for even stateof-the-art language models to solve in single attempts. Search-based evolutionary methods offer a promising alternative by exploring solution spaces iteratively, but their effectiveness remain limited by the fixed capabilities of the underlying generative model. We propose SOAR, a method that learns program synthesis by integrating language models into a self-improving evolutionary loop. SOAR alternates between (1) an evolutionary search that uses an LLM to generate and refine candidate solutions, and (2) a hindsight learning phase that converts search attempts into valid problem-solution pairs used to fine-tune the LLM's generation and refinement capabilities; enabling increasingly effective search in subsequent iterations. On the challenging ARC-AGI benchmark, SOAR achieves significant performance gains across model scales and iterations, leveraging positive transfer between the generation and refinement finetuning tasks. These improvements carry over to test-time adaptation, where SOAR achieves state-of-the-art results among program synthesis methods based on open-source LLMs.

Paperid:155

Authors:Wei Chen · Jun-Xiang Mao · Xiaozheng Wang · Min-Ling Zhang

Title: Learnware Specification via Dual Alignment

Abstract:
The learnware paradigm aims to establish a learnware dock system that contains numerous leanwares, each consisting of a welltrained model and a specification, enabling users to reuse high-performing models for their tasks instead of training from scratch. The specification, as a unique characterization of the model's specialties, dominates the effectiveness of model reuse. Existing specification methods mainly employ distribution alignment to generate specifications. However, this approach overlooks the model's discriminative performance, hindering an adequate specialty characterization. In this paper, we claim that it is beneficial to incorporate such discriminative performance for high-quality specification generation. Accordingly, a novel specification approach named Dali, i.e., Learnware Specification via Dual ALIgnment, is proposed. In Dali, the characterization of the model's discriminative performance is modeled as discriminative alignment, which is considered along with distribution alignment in the specification generation process. Theoretical and empirical analyses clearly demonstrate that the proposed approach is capable of facilitating model reuse in the learnware paradigm with high-quality specification generation.

Paperid:156

Authors:Daniel Eftekhari · Vardan Papyan

Title: On the Importance of Gaussianizing Representations

Abstract:
The normal distribution plays a central role in information theory – it is at the same time the bestcase signal and worst-case noise distribution, has the greatest representational capacity of any distribution, and offers an equivalence between uncorrelatedness and independence for joint distributions. Accounting for the mean and variance of activations throughout the layers of deep neural networks has had a significant effect on facilitating their effective training, but seldom has a prescription for precisely what distribution these activations should take, and how this might be achieved, been offered. Motivated by the information-theoretic properties of the normal distribution, we address this question and concurrently present normality normalization: a novel normalization layer which encourages normality in the feature representations of neural networks using the power transform and employs additive Gaussian noise during training. Our experiments comprehensively demonstrate the effectiveness of normality normalization, in regards to its generalization performance on an array of widely used model and dataset combinations, its strong performance across various common factors of variation such as model width, depth, and training minibatch size, its suitability for usage wherever existing normalization layers are conventionally used, and as a means to improving model robustness to random perturbations.

Paperid:157

Authors:Tiansheng Wen · Yifei Wang · Zequn Zeng · Zhong Peng · Yudi Su · Xinyang Liu · Bo Chen · Hongwei Liu · Stefanie Jegelka · Chenyu You

Title: Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation

Abstract:
Many largescale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show thatsparse codingoffers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We proposeContrastive Sparse Representation(CSR), a method that specifies pre-trained embeddings into a high-dimensional butselectively activatedfeature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed—often by large margins—while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount.

Paperid:158

Authors:Peimeng Guan · Mark Davenport

Title: SGD Jittering: A Training Strategy for Robust and Accurate Model-Based Architectures

Abstract:
Inverse problems aim to reconstruct unseen data from corrupted or perturbed measurements. While most work focuses on improving reconstruction quality, generalization accuracy and robustness are equally important, especially for safetycritical applications. Model-based architectures (MBAs), such as loop unrolling methods, are considered more interpretable and achieve better reconstructions. Empirical evidence suggests that MBAs are more robust to perturbations than black-box solvers, but the accuracy-robustness tradeoff in MBAs remains underexplored. In this work, we propose a simple yet effective training scheme for MBAs, called SGD jittering, which injects noise iteration-wise during reconstruction. We theoretically demonstrate that SGD jittering not only generalizes better than the standard mean squared error training but is also more robust to average-case attacks. We validate SGD jittering using denoising toy examples, seismic deconvolution, and single-coil MRI reconstruction. Both SGD jittering and its SPGD extension yield cleaner reconstructions for out-of-distribution data and demonstrates enhanced robustness against adversarial attacks.

Paperid:159

Authors:Felipe Nuti · Tim Franzmeyer · Joao Henriques

Title: Measuring the Contribution of Fine-Tuning to Individual Responses of LLMs

Abstract:
Abstract:Past work has studied the effects of finetuning on large language models' (LLMs) overall performance on certain tasks. However, a way to quantitatively and systematically analyze its effect on individual outputs is still lacking.In this work, we propose a new method for measuring the contribution that fine-tuning makes to individual LLM responses, assuming access to the original pre-trained model. Our method takes into account the model's intermediate hidden states, giving a more fine-grained insight into the effects of fine-tuning than a simple comparison of the final outputs of pre-trained and fine-tuned models.We introduce and theoretically analyze an exact decomposition of any fine-tuned LLM into a pre-training component and a fine-tuning component.Empirically, we find that one can steer model behavior and performance by up- or down-scaling the fine-tuning component during the forward pass.Motivated by this finding and our theoretical analysis, we define the Tuning Contribution ($\mathrm{TuCo}$) in terms of the ratio of the magnitudes fine-tuning component and the pre-training component.We find that three prominent adversarial attacks on LLMs circumvent safety measures in a way that reduces the Tuning Contribution, and that $\mathrm{TuCo}$ is consistently lower on prompts where the attacks succeed compared to ones where they do not. This suggests that attenuating the effect of fine-tuning on model outputs plays a role in the success of these attacks.In short, $\mathrm{TuCo}$ enables the quantitative study of how fine-tuning influences model behavior and safety, and vice-versa.

Paperid:160

Authors:Jessica Dai · Paula Gradu · Inioluwa Raji · Benjamin Recht

Title: From Individual Experience to Collective Evidence: A Reporting-Based Framework for Identifying Systemic Harms

Abstract:
When an individual reports a negative interaction with some system, how can their personal experience be contextualized within broader patterns of system behavior? We study theincident databaseproblem, where individual reports of adverse events arrive sequentially, and are aggregated over time. In this work, our goal is to identify whether there are subgroups—defined by any combination of relevantfeatures—that are disproportionately likely to experience harmful interactions with the system. We formalize this problem as a sequential hypothesis test, and identify conditions on reporting behavior that are sufficient for making inferences about disparities in true rates of harm across subgroups. We show that algorithms for sequential hypothesis tests can be applied to this problem with a standard multiple testingcorrection. We then demonstrate our method on realworld datasets, including mortgage decisions and vaccine side effects; on each, our method (re-)identifies subgroups known to experience disproportionate harm using only a fraction of the data that was initially used to discover them.

Paperid:161

Authors:Wei Yao · Zeliang Zhang · Huayi Tang · Yong Liu

Title: Understanding Model Ensemble in Transferable Adversarial Attack

Abstract:
Model ensemble adversarial attack has become a powerful method for generating transferable adversarial examples that can target even unknown models, but its theoretical foundation remains underexplored. To address this gap, we provide early theoretical insights that serve as a roadmap for advancing model ensemble adversarial attack. We first define transferability error to measure the error in adversarial transferability, alongside concepts of diversity and empirical model ensemble Rademacher complexity. We then decompose the transferability error into vulnerability, diversity, and a constant, which rigidly explains the origin of transferability error in model ensemble attack: the vulnerability of an adversarial example to ensemble components, and the diversity of ensemble components. Furthermore, we apply the latest mathematical tools in information theory to bound the transferability error using complexity and generalization terms, validating three practical guidelines for reducing transferability error: (1) incorporating more surrogate models, (2) increasing their diversity, and (3) reducing their complexity in cases of overfitting. Finally, extensive experiments with 54 models validate our theoretical framework, representing a significant step forward in understanding transferable model ensemble adversarial attacks.

Paperid:162

Authors:Haowen Ma · Zhiguo Long · Hua Meng

Title: TANGO: Clustering with TypicalityAware Nonlocal Mode-Seeking and Graph-Cut Optimization

Abstract:
Densitybased mode-seeking methods generate a \emph{density-ascending dependency} from low-density points towards higher-density neighbors.Current mode-seeking methods identify modes by breaking some dependency connections, but relying heavily on local data characteristics, requiring case-by-case threshold settings or human intervention to be effective for different datasets.To address this issue, we introduce a novel concept called \emph{typicality}, by exploring the \emph{locally defined} dependency from a \emph{global} perspective, to quantify how confident a point would be a mode. We devise an algorithm that effectively and efficiently identifies modes with the help of the global-view typicality. To implement and validate our idea, we design a clustering method called TANGO, which not only leverages typicality to detect modes, but also utilizes graph-cut with an improved \emph{path-based similarity} to aggregate data into the final clusters. Moreover, this paper also provides some theoretical analysis on the proposed algorithm. Experimental results on several synthetic and extensive real-world datasets demonstrate the effectiveness and superiority of TANGO.

Paperid:163

Authors:Mengyang Sun · Yihao Wang · Tao Feng · Dan Zhang · Yifan Zhu · Jie Tang

Title: A Stronger Mixture of Low-rank Experts for Fine-Tuning Foundation Models

Abstract:
In order to streamline the finetuning of foundation models, Low-Rank Adapters (LoRAs) have been substantially adopted across various fields, including instruction tuning and domain adaptation. The underlying concept of LoRA involves decomposing a full-rank matrix into the product of two lower-rank matrices, which reduces storage consumption and accelerates the training process. Furthermore, to address the limited expressive capacity of LoRA, the Mixture-of-Expert (MoE) has been introduced for incorporating multiple LoRA adapters. The integration of LoRA experts leads to a visible improvement across several downstream scenes. However, the mixture of LoRAs (MoE-LoRA) still exhibits its low robustness during tuning and inferring. Inspired by the Riemannian Preconditioners which train LoRA as a sub-space projector, we propose a new training strategy for MoE-LoRA, to stabilize and boost its feature learning procedure by multi-space projections. Examinations on SGD and AdamW optimizers demonstrate the effectiveness of our methodology. The code will be available upon publication.

Paperid:164

Authors:Peijia Qin · Jianguo Zhang

Title: Learning to Quantize for Training Vector-Quantized Networks

Abstract:
Deep neural networks incorporating discrete latent variables have shown significant potential in sequence modeling.A notable approach is to leverage vector quantization (VQ) to generate discrete representations within a codebook.However, its discrete nature prevents the use of standard backpropagation, which has led to challenges in efficient codebook training.In this work, we introduce MetaQuantization (MQ), a novel vector quantization training framework inspired by meta-learning.Our method separates the optimization of the codebook and the auto-encoder into two levels.Furthermore, we introduce a hyper-net to replace the embedding-parameterized codebook, enabling the codebook to be dynamically generated based on the feedback from the auto-encoder.Different from previous VQ objectives, our innovation results in a meta-objective that makes the codebook training task-aware.We validate the effectiveness of MQ with VQVAE and VQGAN architecture on image reconstruction and generation tasks.Experimental results showcase the superior generative performance of MQ, underscoring its potential as a robust alternative to existing VQ methods.

Paperid:165

Authors:Matthew Joseph · Alex Kulesza · Alexander Yu

Title: Approximate Differential Privacy of the $\ell_2$ Mechanism

Abstract:
Abstract:We study the $\ell_2$ mechanism for computing a $d$dimensional statistic with bounded $\ell_2$ sensitivity under approximate differential privacy. Across a range of privacy parameters, we find that the $\ell_2$ mechanism obtains error approaching that of the Laplace mechanism as $d \to 1$ and approaching that of the Gaussian mechanism as $d \to \infty$; however, it dominates both in between.

Paperid:166

Authors:Yihao Xue · Jiping Li · Baharan Mirzasoleiman

Title: Representations Shape Weak-to-Strong Generalization: Theoretical Insights and Empirical Predictions

Abstract:
Weakto-Strong Generalization (W2SG), where a weak model supervises a stronger one, serves as an important analogy for understanding how humans might guide superhuman intelligence in the future. Promising empirical results revealed that a strong model can surpass its weak supervisor. While recent work has offered theoretical insights into this phenomenon, a clear understanding of the interactions between weak and strong models that drive W2SG remains elusive. We investigate W2SG through a theoretical lens and show that it can be characterized using kernels derived from the principal components of weak and strong models' internal representations. These kernels can be used to define a space that, at a high level, captures what the weak model is unable to learn but is learnable by the strong model. The projection of labels onto this space quantifies how much the strong model falls short of its full potential due to weak supervision. This characterization also provides insights into how certain errors in weak supervision can be corrected by the strong model, regardless of overfitting. Our theory has significant practical implications, providing a representation-based metric that predicts W2SG performance trends without requiring labels, as shown in experiments on molecular predictions with transformers and 5 NLP tasks involving 52 LLMs.

Paperid:167

Authors:Adel Javanmard · Vahab Mirrokni · Jean Pouget-Abadie

Title: Improving the Variance of Differentially Private Randomized Experiments through Clustering

Abstract:
Estimating causal effects from randomized experiments is only possible if participants are willing to disclose their potentially sensitive responses. Differential privacy, a widely used framework for ensuring an algorithm’s privacy guarantees, can encourage participants to share their responses without the risk of deanonymization. However, many mechanisms achieve differential privacy by adding noise to the original dataset, which reduces the precision of causal effect estimation. This introduces a fundamental trade-off between privacy and variance when performing causal analyses on differentially private data.In this work, we propose a new differentially private mechanism, Cluster-DP, which leverages a given cluster structure in the data to improve the privacy-variance trade-off. While our results apply to any clustering, we demonstrate that identifying higher-quality clusters—using a quality measure we define—can reduce the variance penalty while maintaining privacy guarantees. Finally, we evaluate the theoretical and empirical performance of our Cluster-DP algorithm on both real and simulated data, comparing it to common baselines, including two special cases of our algorithm: its unclustered version and a uniform-prior version.

Paperid:168

Authors:Haobo Jiang · Jin Xie · jian Yang · Liang Yu · Jianmin Zheng

Title: Generative Point Cloud Registration

Abstract:
In this paper, we propose a novel 3D registration paradigm, Generative Point Cloud Registration, which bridges advanced 2D generative models with 3D matching tasks to enhance registration performance. Our key idea is to generate crossview consistent image pairs that are well-aligned with the source and target point clouds, enabling geometric-color feature fusion to facilitate robust matching. To ensure high-quality matching, the generated image pair should feature both 2D-3D geometric consistency and cross-view texture consistency. To achieve this, we introduce Match-ControlNet, a matching-specific, controllable 2D generative model. Specifically, it leverages the depth-conditioned generation capability of ControlNet to produce images that are geometrically aligned with depth maps derived from point clouds, ensuring 2D-3D geometric consistency. Additionally, by incorporating a coupled conditional denoising scheme and coupled prompt guidance, Match-ControlNet further promotes cross-view feature interaction, guiding texture consistency generation. Our generative 3D registration paradigm is general and could be seamlessly integrated into various registration methods to enhance their performance. Extensive experiments on 3DMatch and ScanNet datasets verify the effectiveness of our approach.

Paperid:169

Authors:Wei Xiao · Johnson Tsun-Hsuan Wang · Chuang Gan · Daniela Rus

Title: ABNet: Adaptive explicit-Barrier Net for Safe and Scalable Robot Learning

Abstract:
Safe learning is central to AIenabled robots where a single failure may lead to catastrophic results. Existing safe learning methods are not scalable, inefficient and hard to train, and tend to generate unstable signals under noisy inputs that are challenging to be deployed for robots. To address these challenges, we propose Adaptive explicit-Barrier Net (ABNet) in which barriers explicitly show up in the closed-form model that guarantees safety. The ABNet has the potential to incrementally scale toward larger safe foundation models. Each head of ABNet could learn safe control policies from different features and focuses on specific part of the observation. In this way, we do not need to directly construct a large model for complex tasks, which significantly facilitates the training of the model while ensuring its stable output. Most importantly, we can still formally prove the safety guarantees of the ABNet. We demonstrate the efficiency and strength of ABNet in 2D robot obstacle avoidance, safe robot manipulation, and vision-based end-to-end autonomous driving, with results showing much better robustness and guarantees over existing models.

Paperid:170

Authors:Gal Dalal · Assaf Hallak · Gugan Chandrashekhar Mallika Thoppe · Shie Mannor · Gal Chechik

Title: Policy Gradient with Tree Expansion

Abstract:
Policy gradient methods are notorious for having a large variance and high sample complexity. To mitigate this, we introduce SoftTreeMaxa generalization of softmax that employs planning. In SoftTreeMax, we extend the traditional logits with the multi-step discounted cumulative reward, topped with the logits of future states. We analyze SoftTreeMax and explain how tree expansion helps to reduce its gradient variance. We prove that the variance depends on the chosen tree-expansion policy. Specifically, we show that the closer the induced transitions are to being state-independent, the stronger the variance decay. With approximate forward models, we prove that the resulting gradient bias diminishes with the approximation error while retaining the same variance reduction. Ours is the first result to bound the gradient bias for an approximate model. In a practical implementation of SoftTreeMax we utilize a parallel GPU-based simulator for fast and efficient tree expansion. Using this implementation in Atari, we show that SoftTreeMax reduces the gradient variance by three orders of magnitude. This leads to better sample complexity and improved performance compared to distributed PPO.

Paperid:171

Authors:Ignacio Peis · Batuhan Koyuncu · Isabel Valera · Jes Frellsen

Title: Hyper-Transforming Latent Diffusion Models

Abstract:
We introduce a novel generative framework for functions by integrating Implicit Neural Representations (INRs) and Transformerbased hypernetworks into latent variable models. Unlike prior approaches that rely on MLP-based hypernetworks with scalability limitations, our method employs a Transformer-based decoder to generate INR parameters from latent variables, addressing both representation capacity and computational efficiency. Our framework extends latent diffusion models (LDMs) to INR generation by replacing standard decoders with a Transformer-based hypernetwork, which can be trained either from scratch or via hyper-transforming—a strategy that fine-tunes only the decoder while freezing the pre-trained latent space. This enables efficient adaptation of existing generative models to INR-based representations without requiring full retraining. We validate our approach across multiple modalities, demonstrating improved scalability, expressiveness, and generalization over existing INR-based generative models. Our findings establish a unified and flexible framework for learning structured function representations.

Paperid:172

Authors:Anne Ouyang · Simon Guo · Simran Arora · Alex Zhang · William Hu · Christopher Re · Azalia Mirhoseini

Title: KernelBench: Can LLMs Write Efficient GPU Kernels?

Abstract:
Abstract:Efficient GPU kernels are crucial for building performant machine learning architectures, but writing them is a timeconsuming challenge that requires significant expertise; therefore, we explore using language models (LMs) to automate kernel generation. We introduce **KernelBench**, an open-source framework for evaluating LMs' ability to write fast and correct kernels on a suite of 250 carefully selected PyTorch ML workloads. KernelBench represents a real-world engineering environment and making progress on the introduced benchmark directly translates to faster practical kernels. We introduce a new evaluation metric $\text{fast}_p$, which measures the percentage of generated kernels that are functionally correct and offer a speedup greater than an adjustable threshold $p$ over baseline. Our experiments across various state-of-the-art models and test-time methods show that frontier reasoning models perform the best out of the box but still fall short overall, matching the PyTorch baseline in less than 20\% of the cases. While we show that results can improve by leveraging execution and profiling feedback during iterative refinement, KernelBench remains a challenging benchmark, with its difficulty increasing as we raise speedup threshold $p$.

Paperid:173

Authors:Abdelhakim Benechehab · Vasilii Feofanov · Giuseppe Paolo · Albert Thomas · Maurizio Filippone · Balázs Kégl

Title: AdaPTS: Adapting Univariate Foundation Models to Probabilistic Multivariate Time Series Forecasting

Abstract:
Pretrained foundation models (FMs) have shown exceptional performance in univariate time series forecasting tasks. However, several practical challenges persist, including managing intricate dependencies among features and quantifying uncertainty in predictions. This study aims to tackle these critical limitations by introducingadapters—feature-space transformations that facilitate the effective use of pre-trained univariate time series FMs for multivariate tasks. Adapters operate by projecting multivariate inputs into a suitable latent space and applying the FM independently to each dimension in a zero-shot manner. Inspired by the literature on representation learning and partially stochastic Bayesian neural networks, we present a range of adapters and optimization/inference strategies. Experiments conducted on both synthetic and real-world datasets confirm the efficacy of adapters, demonstrating substantial enhancements in forecasting accuracy and uncertainty quantification compared to baseline methods. Our framework,AdaPTS, positions adapters as a modular, scalable, and effective solution for leveraging time series FMs in multivariate contexts, thereby promoting their wider adoption in real-world applications.

Paperid:174

Authors:Shogo Iwazaki · Shion Takeno

Title: Improved Regret Analysis in Gaussian Process Bandits: Optimality for Noiseless Reward, RKHS norm, and Non-Stationary Variance

Abstract:
We study the Gaussian process (GP) bandit problem, whose goal is to minimize regret under an unknown reward function lying in some reproducing kernel Hilbert space (RKHS). The maximum posterior variance analysis is vital in analyzing nearoptimal GP bandit algorithms such as maximum variance reduction (MVR) and phased elimination (PE).Therefore, we first show the new upper bound of the maximum posterior variance, which improves the dependence of the noise variance parameters of the GP. By leveraging this result, we refine the MVR and PE to obtain (i) a nearly optimal regret upper bound in the noiseless setting and (ii) regret upper bounds that are optimal with respect to the RKHS norm of the reward function. Furthermore, as another application of our proposed bound, we analyze the GP bandit under the time-varying noise variance setting, which is the kernelized extension of the linear bandit with heteroscedastic noise. For this problem, we show that MVR and PE-based algorithms achieve noise variance-dependent regret upper bounds, which matches our regret lower bound.

Paperid:175

Authors:Hongwei Li · Yuheng Tang · Shiqi Wang · Wenbo Guo

Title: PatchPilot: A Stable and Cost-Efficient Agentic Patching Framework

Abstract:
Recent research builds various patching agents that combine large language models (LLMs) with nonML tools and achieve promising results on the state-of-the-art (SOTA) software patching benchmark, SWE-Bench. Based on how to determine the patching workflows, existing patching agents can be categorized as agent-based planning methods, which rely on LLMs for planning, and human-based planning methods, which follow a pre-defined workflow.At a high level, agent-based planning methods achieve high patching performance but with a high cost and limited stability. Human-based planning methods, on the other hand, are more stable and efficient but have key workflow limitations that compromise their patching performance.In this paper, we propose PatchPilot, an agentic patcher that strikes a balance between patching efficacy, stability, and cost-efficiency. PatchPilot proposes a novel human-based planning workflow with five components: reproduction, localization, generation, validation, and refinement (where refinement is unique to PatchPilot).We introduce novel and customized designs to each component to optimize their effectiveness and efficiency. Through extensive experiments on the SWE-Bench benchmarks, PatchPilot shows a superior performance than existing open-source methods while maintaining low cost (less than 1\$ per instance) and ensuring higher stability.We also conduct a detailed ablation study to validate the key designs in each component.

Paperid:176

Authors:Bariscan Kurtkaya · Fatih Dinc · Mert Yuksekgonul · Marta Blanco-Pozo · Ege Cirakman · Mark Schnitzer · Yucel Yemez · Hidenori Tanaka · Yuan · Nina Miolane

Title: Dynamical phases of short-term memory mechanisms in RNNs

Abstract:
Shortterm memory is essential for cognitive processing, yet our understanding of its neural mechanisms remains unclear. A key focus in neuroscience has been the study of sequential activity patterns, where neurons fire one after another within large networks, to explain how information is maintained. While recurrent connections were shown to drive sequential dynamics, a mechanistic understanding of this process still remains unknown. In this work, we first introduce two unique mechanisms that can subserve short-term memory: slow-point manifolds generating direct sequences or limit cycles providing temporally localized approximations. Then, through analytical models, we identify fundamental properties that govern the selection of these mechanisms, \textit{i.e.}, we derive theoretical scaling laws for critical learning rates as a function of the delay period length, beyond which no learning is possible. We empirically verify these observations by training and evaluating more than 35,000 recurrent neural networks (RNNs) that we will publicly release upon publication. Overall, our work provides new insights into short-term memory mechanisms and proposes experimentally testable predictions for systems neuroscience.

Paperid:177

Authors:Lutfi Erdogan · Hiroki Furuta · Sehoon Kim · Nicholas Lee · Suhong Moon · Gopala Anumanchipalli · Kurt Keutzer · Amir Gholaminejad

Title: Plan-and-Act: Structured Execution of LLM Agents for Long-Horizon Web Tasks

Abstract:
Large language models (LLMs) have shown remarkable advancements in enabling language agents to tackle complex, multistep, long-horizon tasks such as web navigation. However, existing approaches struggle with challenges like misalignment between high-level user queries and low-level executable actions, difficulty in maintaining consistency over long-horizon tasks, and a lack of grounding in specific environments. To address these issues, we propose Plan-and-Act, a novel framework that incorporates explicit planning into LLM-based agents. Plan-and-Act consists of a Planner model that generates high-level plans to achieve user goals and an Executor model that translates these plans into precise, environment-specific actions. By explicitly separating planning from execution, Plan-and-Act improves alignment between the high-level reasoning and low-level actions, which enables consistent decision-making and adaptability to changes in the environment. To train the Planner effectively, we introduce a synthetic data generation method that annotates ground-truth trajectories with feasible plans, augmented by diverse and extensive examples to enhance generalization. We evaluate Plan-and-Act in the WebArena environment, demonstrating a SOTA 54% success rate on the WebArena-Lite benchmark.

Paperid:178

Authors:Sam Gijsen · Kerstin Ritter

Title: EEG-Language Pretraining for Highly Label-Efficient Pathology Detection

Abstract:
Multimodal language modeling has enabled breakthroughs for representation learning, yet remains unexplored in the realm of functional brain data for pathology detection. This paper pioneers EEGlanguage models (ELMs) trained on clinical reports and 15000 EEGs. We propose to combine multimodal alignment in this novel domain with timeseries cropping and text segmentation, enabling an extension based on multiple instance learning to alleviate misalignment between irrelevant EEG or text segments. Our multimodal models significantly improve pathology detection compared to EEG-only models across four evaluations and for the first time enable zero-shot classification as well as retrieval of both neural signals and reports. In sum, these results highlight the potential of ELMs, representing significant progress for clinical applications.

Paperid:179

Authors:Yinyan Bu · Jiajie Yu · Kai Zheng · Xinyu Zhang · Piya Pal

Title: NEAR: Neural Electromagnetic Array Response

Abstract:
We address the challenge of achieving angular superresolution in multi-antenna radar systems that are widely used for localization, navigation, and automotive perception. A multi-antenna radar achieves very high resolution by computationally creating a large virtual sensing system using very few physical antennas. However, practical constraints imposed by hardware, noise, and a limited number of antennas can impede its performance. Conventional supervised learning models that rely on extensive pre-training with large datasets, often exhibit poor generalization in unseen environments. To overcome these limitations, we propose NEAR, an untrained implicit neural representation (INR) framework that predicts radar responses at unseen locations from sparse measurements, by leveraging latent harmonic structures inherent in radar wave propagation. We establish new theoretical results linking antenna array response to expressive power of INR architectures, and develop a novel physics-informed and latent geometry-aware regularizer. Our approach integrates classical signal representation with modern implicit neural learning, enabling high-resolution radar sensing that is both interpretable and generalizable. Extensive simulations and real-world experiments using radar platforms demonstrate NEAR's effectiveness and its ability to adapt to unseen environments.

Paperid:180

Authors:Wei Qu · Jiawei Guan · Rui Ma · kezhai · weikun wu · haobo Wang

Title: P(all-atom) Is Unlocking New Path For Protein Design

Abstract:
Abstract:We introduce Pallatom, an innovative protein generation model capable of producing protein structures with allatom coordinates. Pallatom directly learns and models the joint distribution $P(\textit{structure}, \textit{seq})$ by focusing on $P(\textit{all-atom})$, effectively addressing the interdependence between sequence and structure in protein generation. To achieve this, we propose a novel network architecture specifically designed for all-atom protein generation. Our model employs a dual-track framework that tokenizes proteins into token-level and atomic-level representations, integrating them through a multi-layer decoding process with "traversing" representations and recycling mechanism. We also introduce the $\texttt{atom14}$ representation method, which unifies the description of unknown side-chain coordinates, ensuring high fidelity between the generated all-atom conformation and its physical structure. Experimental results demonstrate that Pallatom excels in key metrics of protein design, including designability, diversity, and novelty, showing significant improvements across the board. Our model not only enhances the accuracy of protein generation but also exhibits excellent training efficiency, paving the way for future applications in larger and more complex systems.

Paperid:181

Authors:Yen-Jung Hsu · Ming-Yi Hong · Miao-Chen Chiang · Che Lin

Title: MTSTRec: Multimodal Time-Aligned Shared Token Recommender

Abstract:
Sequential recommendation in ecommerce utilizes users' anonymous browsing histories to personalize product suggestions without relying on private information. Existing item ID-based methods and multimodal models often overlook the temporal alignment of modalities like textual descriptions, visual content, and prices in user browsing sequences. To address this limitation, this paper proposes the Multimodal Time-aligned Shared Token Recommender (MTSTRec), a transformer-based framework with a single time-aligned shared token per product for efficient cross-modality fusion. MTSTRec preserves the distinct contributions of each modality while aligning them temporally to better capture user preferences. Extensive experiments demonstrate that MTSTRec achieves state-of-the-art performance across multiple sequential recommendation benchmarks, significantly improving upon existing multimodal fusion.

Paperid:182

Authors:Kai Liu · Kaicheng Yang · Zheng Chen · Zhiteng Li · Yong Guo · Wenbo Li · Linghe Kong · Yulun Zhang

Title: BiMaCoSR: Binary One-Step Diffusion Model Leveraging Flexible Matrix Compression for Real Super-Resolution

Abstract:
While superresolution (SR) methods based on diffusion models (DM) have demonstrated inspiring performance, their deployment is impeded due to the heavy request of memory and computation. Recent researchers apply two kinds of methods to compress or fasten the DM. One is to compress the DM into 1-bit, aka binarization, alleviating the storage and computation pressure. The other distills the multi-step DM into only one step, significantly speeding up inference process. Nonetheless, it remains impossible to deploy DM to resource-limited edge devices. To address this problem, we propose BiMaCoSR, which combines binarization and one-step distillation to obtain extreme compression and acceleration. To prevent the catastrophic collapse of the model caused by binarization, we proposed sparse matrix branch (SMB) and low rank matrixbranch (LRM). Both auxiliary branches pass the full-precision (FP) information but in different ways. SMB absorbs the extreme values and its output is high rank, carrying abundant FP information. Whereas, the design of LRMB is inspired by LoRA and is initialized with the top r SVD components, outputting low rank representation. The computation and storage overhead of our proposed branches can be safely ignored. Comprehensive comparison experiments are conducted to exhibit BiMaCoSR outperforms current state-of-the-art binarization methods and gains competitive performance compared with FP one-step model. BiMaCoSR achieves a 23.8x compression ratio and a 27.4x speedup ratio compared to FP counterpart. Our code will be released soon.

Paperid:183

Authors:Jingyang Ke · Feiyang Wu · Jiyi Wang · Jeffrey Markowitz · Anqi Wu

Title: Inverse Reinforcement Learning with Switching Rewards and History Dependency for Characterizing Animal Behaviors

Abstract:
Traditional approaches to studying decisionmaking in neuroscience focus on simplified behavioral tasks where animals perform repetitive, stereotyped actions to receive explicit rewards. While informative, these methods constrain our understanding of decision-making to short timescale behaviors driven by explicit goals. In natural environments, animals exhibit more complex, long-term behaviors driven by intrinsic motivations that are often unobservable. Recent works in time-varying inverse reinforcement learning (IRL) aim to capture shifting motivations in long-term, freely moving behaviors. However, a crucial challenge remains: animals make decisions based on their history, not just their current state. To address this, we introduce SWIRL (SWitching IRL), a novel framework that extends traditional IRL by incorporating time-varying, history-dependent reward functions. SWIRL models long behavioral sequences as transitions between short-term decision-making processes, each governed by a unique reward function. SWIRL incorporates biologically plausible history dependency to capture how past decisions and environmental contexts shape behavior, offering a more accurate description of animal decision-making. We apply SWIRL to simulated and real-world animal behavior datasets and show that it outperforms models lacking history dependency, both quantitatively and qualitatively. This work presents the first IRL model to incorporate history-dependent policies and rewards to advance our understanding of complex, naturalistic decision-making in animals.

Paperid:184

Authors:Ethan Rathbun · Christopher Amato · Alina Oprea

Title: Adversarial Inception Backdoor Attacks against Reinforcement Learning

Abstract:
Recent works have demonstrated the vulnerability of Deep Reinforcement Learning (DRL) algorithms against trainingtime, backdoor poisoning attacks. The objectives of these attacks are twofold: induce pre-determined, adversarial behavior in the agent upon observing a fixed trigger during deployment while allowing the agent to solve its intended task during training. Prior attacks assume arbitrary control over the agent's rewards, inducing values far outside the environment's natural constraints. This results in brittle attacks that fail once the proper reward constraints are enforced. Thus, in this work we propose a new class of backdoor attacks against DRL which are the first to achieve state of the art performance under strict reward constraints. These ``inception'' attacks manipulate the agent's training data -- inserting the trigger into prior observations and replacing high return actions with those of the targeted adversarial behavior. We formally define these attacks and prove they achieve both adversarial objectives against arbitrary Markov Decision Processes (MDP). Using this framework we devise an online inception attack which achieves an 100% attack success rate on multiple environments under constrained rewards while minimally impacting the agent's task performance.

Paperid:185

Authors:Zhaohe Liao · Jiangtong Li · Siyu Sun · Qingyang Liu · Fengshun Xiao · Tianjiao Li · Qiang Zhang · Guang Chen · Li Niu · Changjun Jiang · Liqing Zhang

Title: Divide and Conquer: Exploring Language-centric Tree Reasoning for Video Question-Answering

Abstract:
Video QuestionAnswering (VideoQA) remains challenging in achieving advanced cognitive reasoning due to the uncontrollable and opaque reasoning processes in existing Multimodal Large Language Models (MLLMs). To address this issue, we propose a novel Language-centric Tree Reasoning (LTR) framework that targets on enhancing the reasoning ability of models. In detail, it recursively divides the original question into logically manageable parts and conquers them piece by piece, enhancing the reasoning capabilities and interpretability of existing MLLMs. Specifically, in the first stage, the LTR focuses on language to recursively generate a language-centric logical tree, which gradually breaks down the complex cognitive question into simple perceptual ones and plans the reasoning path through a RAG-based few-shot approach. In the second stage, with the aid of video content, the LTR performs bottom-up logical reasoning within the tree to derive the final answer along with the traceable reasoning path. Experiments across 11 VideoQA benchmarks demonstrate that our LTR framework significantly improves both accuracy and interpretability compared to state-of-the-art MLLMs. To our knowledge, this is the first work to implement a language-centric logical tree to guide MLLM reasoning in VideoQA, paving the way for language-centric video understanding from perception to cognition.

Paperid:186

Authors:Steve Hanneke · Qinglin Meng · Amirreza Shaeiri

Title: Representation Preserving Multiclass Agnostic to Realizable Reduction

Abstract:
We study the problem of multiclass classification when the number of labels can be unbounded within the PAC learning framework. Our main contribution is a theory that demonstrates asimpleandelegantagnostic to realizable reduction for this framework. This resolves an open problem raised by the recent work of (Hopkins et al., 2022). Notably, our result is the firstrepresentation preservingmulticlass agnostic to realizable reduction, in contrast with the compression based approach of the work of (David et al., 2017). Furthermore, our main theorem is stated in an abstract framework, called ``Unified PAC Learning'', which encompasses a range of frameworks, including multiclass PAC learning, list PAC learning, and multilabel PAC learning. In addition, we explore representation preserving reductions to the realizable setting for two noise models, namely Massart noise and Tsybakov noise, in the multiclass PAC learning framework. We believe our technique may find other applications in ensuing studies of theoretical machine learning.

Paperid:187

Authors:Tuan Truong · Ngọc Quân Phạm · Quyen Tran · Tan Nguyen · Dinh Phung · Trung Le

Title: Promoting Ensemble Diversity with Interactive Bayesian Distributional Robustness for Finetuning Foundation Models

Abstract:
We introduce Interactive Bayesian Distributional Robustness (IBDR), a novel Bayesian inference framework that allows modeling the interactions between particles, thereby enhancing ensemble quality through increased particle diversity. IBDR is grounded in a generalized theoretical framework that connects the distributional population loss with the approximate posterior, motivating a practical dual optimization procedure that enforces distributional robustness while fostering particle diversity. We evaluate IBDR's performance against various baseline methods using the VTAB1K benchmark and the common reasoning language task. The results consistently show that IBDR outperforms these baselines, underscoring its effectiveness in real-world applications. Our code is publicly available at https://anonymous.4open.science/r/DRO-1C12/

Paperid:188

Authors:Ruiquan Huang · Yingbin LIANG · Jing Yang

Title: How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias

Abstract:
Abstract:Language recognition tasks are fundamental in natural language processing (NLP) and have been widely used to benchmark the performance of large language models (LLMs). These tasks also play a crucial role in explaining the working mechanisms of transformers. In this work, we focus on two representative tasks in the category of regular language recognition, known as 'even pairs' and 'parity check', the aim of which is to determine whether the occurrences of certain subsequences in a given sequence are even. Our goal is to explore how a onelayer transformer, consisting of an attention layer followed by a linear layer, learns to solve these tasks by theoretically analyzing its training dynamics under gradient descent. While even pairs can be solved directly by a one-layer transformer, parity check need to be solved by integrating Chain-of-Thought (CoT), either into the inference stage of a transformer well-trained for the even pairs task, or into the training of a one-layer transformer. For both problems, our analysis shows that the joint training of attention and linear layers exhibits two distinct phases. In the first phase, the attention layer grows rapidly, mapping data sequences into separable vectors. In the second phase, the attention layer becomes stable, while the linear layer grows logarithmically and approaches in direction to a max-margin hyperplane that correctly separates the attention layer outputs into positive and negative samples, and the loss decreases at a rate of $O(1/t)$. Our experiments validate those theoretical results.

Paperid:189

Authors:Xu Wan · Wenyue Xu · Chao Yang · Mingyang Sun

Title: Think Twice, Act Once: A Co-Evolution Framework of LLM and RL for Large-Scale Decision Making

Abstract:
Recent advancements in Large Language Models (LLMs) and Reinforcement Learning (RL) have shown significant promise in decisionmaking tasks. Nevertheless, for large-scale industrial decision problems, both approaches face distinct challenges: LLMs lack real-time long-sequence decision-making capabilities, while RL struggles with sample efficiency in vast action spaces. To bridge this gap, we proposeAgentsCo-Evolution (ACE), a synergistic framework between LLMs and RL agent for large-scale decision-making scenarios. ACE introduces a dual-role trajectory refinement mechanism where LLMs act as both Policy Actor and Value Critic during RL's training: the Actor refines suboptimal actions via multi-step reasoning and environment validation, while the Critic performs temporal credit assignment through trajectory-level reward shaping. Concurrently, RL agent enhance LLMs' task-specific decision-making via prioritized experience replay.Through extensive experiments across multiple power grid operation challenges with action spaces exceeding 60K discrete actions, ACE demonstrates superior performance over existing RL methods and LLM-based methods.

Paperid:190

Authors:Johannes Hertrich · Robert Gruhlke

Title: Importance Corrected Neural JKO Sampling

Abstract:
In order to sample from an unnormalized probability density function, we propose to combine continuous normalizing flows (CNFs) with rejectionresampling steps based on importance weights. We relate the iterative training of CNFs with regularized velocity fields to a JKO scheme and prove convergence of the involved velocity fields to the velocity field of the Wasserstein gradient flow (WGF). The alternation of local flow steps and non-local rejection-resampling steps allows to overcome local minima or slow convergence of the WGF for multimodal distributions. Since the proposal of the rejection step is generated by the model itself, they do not suffer from common drawbacks of classical rejection schemes. The arising model can be trained iteratively, reduces the reverse Kullback-Leibler (KL) loss function in each step, allows to generate iid samples and moreover allows for evaluations of the generated underlying density. Numerical examples show that our method yields accurate results on various test distributions including high-dimensional multimodal targets and outperforms the state of the art in almost all cases significantly.

Paperid:191

Authors:Guangyi Wang · Wei Peng · lijiang Li · Wenyu Chen · Yuren Cai · Song-Zhi Su

Title: Diffusion Sampling Correction via Approximately 10 Parameters

Abstract:
While powerful for generation, Diffusion Probabilistic Models (DPMs) face slow sampling challenges, for which various distillationbased methods have been proposed. However, they typically require significant additional training costs and model parameter storage, limiting their practicality. In this work, we propose PCA-based Adaptive Search (PAS), which optimizes existing solvers for DPMs with minimal additional costs. Specifically, we first employ PCA to obtain a few basis vectors to span the high-dimensional sampling space, which enables us to learn just a set of coordinates to correct the sampling direction; furthermore, based on the observation that the cumulative truncation error exhibits an ``S''-shape, we design an adaptive search strategy that further enhances the sampling efficiency and reduces the number of stored parameters to approximately 10. Extensive experiments demonstrate that PAS can significantly enhance existing fast solvers in a plug-and-play manner with negligible costs. E.g., on CIFAR10, PAS optimizes DDIM's FID from 15.69 to 4.37 (NFE=10) using only 12 parameters and sub-minute training on a single A100 GPU. Code will be available.

Paperid:192

Authors:Marco Cusumano-Towner · David Hafner · Alexander Hertzberg · Brody Huval · Aleksei Petrenko · Eugene Vinitsky · Erik Wijmans · Taylor Killian · Stuart Bowers · Ozan Sener · Philipp Kraehenbuehl · Vladlen Koltun

Title: Robust Autonomy Emerges from Self-Play

Abstract:
Selfplay has powered breakthroughs in two-player and multi-player games. Here we show that self-play is a surprisingly effective strategy in another domain. We show that robust and naturalistic driving emerges entirely from self-play in simulation at unprecedented scale -- 1.6 billion km of driving. This is enabled by Gigaflow, a batched simulator that can synthesize and train on 42 years of subjective driving experience per hour on a single 8-GPU node. The resulting policy achieves state-of-the-art performance on three independent autonomous driving benchmarks. The policy outperforms the prior state of the art when tested on recorded real-world scenarios, amidst human drivers, without ever seeing human data during training. The policy is realistic when assessed against human references and achieves unprecedented robustness, averaging 17.5 years of continuous driving between incidents in simulation.

Paperid:193

Authors:Aojun Lu · Tao Feng · Hangjie Yuan · Yanan Sun

Title: Rethinking the Stability-Plasticity Trade-off in Continual Learning from an Architectural Perspective

Abstract:
The quest for Continual Learning (CL) seeks to empower neural networks with the ability to learn and adapt incrementally. Central to this pursuit is addressing the stabilityplasticity dilemma, which involves striking a balance between two conflicting objectives: preserving previously learned knowledge and acquiring new knowledge. Existing studies have proposed numerous CL methods to achieve this trade-off. However, these methods often overlook the impact of basic architecture on stability and plasticity, thus the trade-off is limited to the parameter level. In this paper, we delve into the conflict between stability and plasticity at the architectural level. We reveal that under an equal parameter constraint, deeper networks exhibit better plasticity, while wider networks are characterized by superior stability. To address this architectural-level dilemma, we introduce a novel framework denoted Dual-Architecture (Dual-Arch), which serves as a plug-in component for CL. This framework leverages the complementary strengths of two distinct and independent networks: one dedicated to plasticity and the other to stability. Each network is designed with a specialized and lightweight architecture, tailored to its respective objective. Extensive experiments across datasets and CL methods demonstrate that Dual-Arch can enhance the performance of existing CL methods while being up to87\%more compact in terms of parameters than the baselines.

Paperid:194

Authors:Nearchos Potamitis · Lars Klein · Roland Aydin · Robert West · Caglar Gulcehre · Akhil Arora

Title: Fleet of Agents: Coordinated Problem Solving with Large Language Models

Abstract:
Abstract:While numerous frameworks have been developed to enhance the reasoning abilities of large language models (LLMs), there is a scarcity of methods that effectively balance the tradeoff between cost and quality. In this paper, we introduce Fleet of Agents (FoA), a novel and intuitive yet principled framework utilizing LLMs as agents to navigate through dynamic tree searches, employing a genetic-type particle filtering approach. FoA spawns a multitude of agents, each exploring the search space autonomously, followed by a selection phase where resampling based on a heuristic value function optimizes the balance between exploration and exploitation. This mechanism enables dynamic branching, adapting the exploration strategy based on discovered solutions. We conduct extensive experiments on three benchmark tasks, “Game of 24”, “Mini-Crosswords”, and “WebShop”, utilizing four different LLMs, “GPT-3.5”, “GPT-4”, “LLaMA3.2-11B”, and “LLaMA3.2-90B”. On average across all tasks and LLMs, FoA obtains a quality improvement of $\simeq5$% while requiring only $\simeq40$% of the cost of previous SOTA methods. Notably, our analyses reveal that (1) FoA achieves the best cost-quality trade-off among all benchmarked methods and (2) FoA + LLaMA3.2-11B surpasses the Llama3.2-90B model. FoA is publicly available at https://anonymous.4open.science/r/FoA-4D83

Paperid:195

Authors:Woohyun Lee · Hogun Park

Title: Self-supervised Adversarial Purification for Graph Neural Networks

Abstract:
Defending Graph Neural Networks (GNNs) against adversarial attacks requires balancing accuracy and robustness, a tradeoff often mishandled by traditional methods like adversarial training that intertwine these conflicting objectives within a single classifier. To overcome this limitation, we propose a self-supervised adversarial purification framework. We separate robustness from the classifier by introducing a dedicated purifier, which cleanses the input data before classification. In contrast to prior adversarial purification methods, we propose GPR-GAE, a novel graph auto-encoder, as a specialized purifier trained with a self-supervised strategy, adapting to diverse graph structures in a data-driven manner. Utilizing multiple Generalized PageRank (GPR) filters, GPR-GAE captures diverse structural representations for robust and effective purification. Our multi-step purification process further facilitates GPR-GAE to achieve precise graph recovery and robust defense against structural perturbations. Experiments across diverse datasets and attack scenarios demonstrate state-of-the-art robustness of GPR-GAE, showcasing it as an independent plug-and-play purifier for GNN classifiers.

Paperid:196

Authors:Hila Chefer · Uriel Singer · Amit Zohar · Yuval Kirstain · Adam Polyak · Yaniv Taigman · Lior Wolf · Shelly Sheynin

Title: VideoJAM: Joint Appearance-Motion Representations for Enhanced Motion Generation in Video Models

Abstract:
Despite tremendous recent progress, generative video models still struggle to capture realworld motion, dynamics, and physics. We show that this limitation arises from the conventional pixel reconstruction objective, which biases models toward appearance fidelity at the expense of motion coherence.To address this, we introduceVideoJAM, a novel framework that instills an effective motion prior to video generators, by encouraging the model to learna joint appearance-motion representation. VideoJAM is composed of two complementary units. During training, we extend the objective to predict both the generated pixels and their corresponding motion from a single learned representation. During inference, we introduceInner-Guidance, a mechanism that steers the generation toward coherent motion by leveraging the model's own evolving motion prediction as a dynamic guidance signal.Notably, our framework can be applied to any video model with minimal adaptations, requiring no modifications to the training data or scaling of the model.VideoJAM achieves state-of-the-art performance in motion coherence, surpassing highly competitive proprietary models while also enhancing the perceived visual quality of the generations.These findings emphasize that appearance and motion can be complementary and, when effectively integrated, enhance both the visual quality and the coherence of video generation.

Paperid:197

Authors:Ostmeier · Brian Axelrod · Maya Varma · Michael Moseley · Akshay Chaudhari · Curtis Langlotz

Title: LieRE: Lie Rotational Positional Encodings

Abstract:
Transformer architectures rely on position encodings to capture token dependencies. Rotary Position Encoding (RoPE) has emerged as a popular choice in language models due to its efficient encoding of relative position information through keyquery rotations. However, RoPE faces significant limitations beyond language processing: it is constrained to one-dimensional sequence data and, even with learnable phases, offers limited representational capacity.We address these challenges with Lie Relative Encodings (LieRE), which replaces RoPE's block-2D rotation matrix with a learned, dense, high-dimensional rotation matrix of variable density. Through extensive evaluation on three image datasets across 2D and 3D classification tasks, LieRE achieves 2\% relative improvement over state-of-the-art baselines on 2D tasks and 1.5\% on 3D tasks, while demonstrating superior generalization to higher resolutions. Our implementation is computationally efficient, with results reproducible on 4 A100 GPUs in 30 minutes on CIFAR100, and we release our code to facilitate further research.

Paperid:198

Authors:Ian Gemp · Andreas Haupt · Luke Marris · Siqi Liu · Georgios Piliouras

Title: Convex Markov Games: A New Frontier for Multi-Agent Reinforcement Learning

Abstract:
Behavioral diversity, expert imitation, fairness, safety goals and others give rise to preferences in sequential decision making domains that do not decompose additively across time. We introduce the class of convex Markov games that allow general convex preferences over occupancy measures. Despite infinite time horizon and strictly higher generality than Markov games, pure strategy Nash equilibria exist. Furthermore, equilibria can be approximated empirically by performing gradient descent on an upper bound of exploitability. Our experiments reveal novel solutions to classic repeated normalform games, find fair solutions in a repeated asymmetric coordination game, and prioritize safe long-term behavior in a robot warehouse environment. In the prisoner's dilemma, our algorithm leverages transient imitation to find a policy profile that deviates from observed human play only slightly, yet achieves higher per-player utility while also being three orders of magnitude less exploitable.

Paperid:199

Authors:Nikita Tsoy · Ivan Kirev · Negin Rahimiyazdi · Nikola Konstantinov

Title: On the Impact of Performative Risk Minimization for Binary Random Variables

Abstract:
Performativity, the phenomenon where outcomes are influenced by predictions, is particularly prevalent in social contexts where individuals strategically respond to a deployed model. In order to preserve the high accuracy of machine learning models under distribution shifts caused by performativity, Perdomo et al. (2020) introduced the concept of performative risk minimization (PRM). While this framework ensures model accuracy, it overlooks the impact of the PRM on the underlying distributions and the predictions of the model. In this paper, we initiate the analysis of the impact of PRM, by studying performativity for a sequential performative risk minimization problem with binary random variables and linear performative shifts. We formulate two natural measures of impact. In the case of full information, where the distribution dynamics are known, we derive explicit formulas for the PRM solution and our impact measures. In the case of partial information, we provide performativeaware statistical estimators, as well as simulations. Our analysis contrasts PRM to alternatives that do not model data shift and indicates that PRM can have amplified side effects compared to such methods.

Paperid:200

Authors:Junyan Liu · Lillian Ratliff

Title: Principal-Agent Bandit Games with Self-Interested and Exploratory Learning Agents

Abstract:
Abstract:We study the repeated principalagent bandit game, where the principal indirectly explores an unknown environment by incentivizing an agent to play arms. Unlike prior work that assumes a greedy agent with full knowledge of reward means, we consider a self-interested learning agent who iteratively updates reward estimates and may explore arbitrarily with some probability.As a warm-up, we first consider a self-interested learning agent without exploration.We propose algorithms for both i.i.d. and linear reward settings with bandit feedback in a finite horizon $T$, achieving regret bounds of $\widetilde{O}(\sqrt{T})$ and $\widetilde{O}(T^{\frac{2}{3}})$, respectively. Specifically, these algorithms rely on a novel elimination framework coupled with new search algorithms which accommodate the uncertainty from the agent's learning behavior. We then extend the framework to handle an exploratory learning agent and develop an algorithm to achieve a $\widetilde{O}(T^{\frac{2}{3}})$ regret bound in i.i.d. reward setup by enhancing the robustness of our elimination framework to the potential agent exploration. Finally, when our agent model reduces to that in (Dogan et al., 2023a), we propose an algorithm based on our robust framework, which achieves a $\widetilde{O}(\sqrt{T})$ regret bound, significantly improving upon their $\widetilde{O}(T^{\frac{11}{12}})$ bound.

Paperid:201

Authors:Irmak Sivgin · Sara Fridovich-Keil · Gordon Wetzstein · Mert Pilanci

Title: Geometric Algebra Planes: Convex Implicit Neural Volumes

Abstract:
Volume parameterizations abound in recent literature, encompassing methods from classic voxel grids to implicit neural representations. While implicit representations offer impressive capacity and improved memory efficiency compared to voxel grids, they traditionally require training through nonconvex optimization, which can be slow and sensitive to initialization and hyperparameters. We introduce GAPlanes, a novel family of implicit neural volume representations inspired by Geometric Algebra that can be trained using convex optimization, addressing the limitations of nonconvex methods. GA-Planes models generalize many existing representations including any combination of features stored in tensor basis elements followed by a neural feature decoder, and can be adapted to convex or nonconvex training as needed for various inverse problems. In the 2D setting, we prove GA-Planes models are equivalent to a low-rank plus low-resolution matrix factorization that outperforms the classic low-rank plus sparse decomposition for fitting a natural image. In 3D, GA-Planes models exhibit competitive expressiveness, model size, and optimizability across tasks such as radiance field reconstruction, 3D segmentation, and video segmentation.

Paperid:202

Authors:Chenxi Wang · Linxiao Yang · Zhixian Wang · Liang Sun · Yi Wang

Title: A Non-isotropic Time Series Diffusion Model with Moving Average Transitions

Abstract:
Diffusion models, known for their generative ability, have recently been adapted to time series analysis. Most pioneering works rely on the standard isotropic diffusion, treating each time step and the entire frequency spectrum identically. However, it may not be suitable for time series, which often have more informative lowfrequency components. We empirically found that direct application of standard diffusion to time series may cause gradient contradiction during training, due to the rapid decrease of low-frequency information in the diffusion process. To this end, we proposed a novel time series diffusion model, MA-TSD, which utilizes the moving average, a natural low-frequency filter, as the forward transition. Its backward process is accelerable like DDIM and can be further considered a time series super-resolution. Our experiments on various datasets demonstrated MA-TSD's superior performance in time series forecasting and super-resolution tasks.

Paperid:203

Authors:Lorenzo Lucchese · Mikko Pakkanen · Almut E. D. Veraart

Title: Learning with Expected Signatures: Theory and Applications

Abstract:
The expected signature maps a collection of data streams to a lower dimensional representation, with a remarkable property: the resulting feature tensor can fully characterize the data generating distribution. This "modelfree" embedding has been successfully leveraged to build multiple domain-agnostic machine learning (ML) algorithms for time series and sequential data. The convergence results proved in this paper bridge the gap between the expected signature's empirical discrete-time estimator and its theoretical continuous-time value, allowing for a more complete probabilistic interpretation of expected signature-based ML methods. Moreover, when the data generating process is a martingale, we suggest a simple modification of the expected signature estimator with significantly lower mean squared error and empirically demonstrate how it can be effectively applied to improve predictive performance.

Paperid:204

Authors:junmin zhong · Emiliano Quinones Yumbla · Seyed Yousef Soltanian · Ruofan Wu · Wenlong Zhang · Jennie Si

Title: Reinforcement Learning Control of a Physical Robot Device for Assisted Human Level Ground Walking without a Simulator

Abstract:
This study proposes an innovative approach to soft exosuitassisted human walking. Inspired by this real-life problem involving a wearable robot, we seek to overcome persistent difficulties in ensuring reliable reinforcement learning (RL)-based methods for physical device control. To address the multitude of challenges including limited data and the lack of a simulator engine for human-robot interaction duringwalking, the demand of reduced computational overhead for real-time deployment,and time-efficient adaptation to achieve personalized control while ensuring human safety, we propose an online Adaptation from an offline Imitating expert Policy (AIP) approach. Our offline learning mimics human expert actions through real human walking demonstrations without robot assistance. The resulted policy is then used to initialize online actor-critic learning, the goal of which is to optimally personalize robot assistance. In addition to being fast and robust, our online RL method also posses important properties such as learning convergence, dynamic stability, and solution optimality. We have successfully demonstrated our simple and robust framework for safe robot control on all five tested human participants,without selectively presenting results. The qualitative performance guarantees provided by our online RL, along with the consistent experimental validation of AIP control, represent the first demonstration of online adaptation for softsuit control personalization and serve as important evidence for the use of online RL in controlling a physical device to solve a real-life problem.

Paperid:205

Authors:Yu Zhou · Bingxuan Li · Mohan Tang · Xiaomeng Jin · Te-Lin Wu · Kuan-Hao Huang · Heng Ji · Kai-Wei Chang · Nanyun Peng

Title: Contrastive Visual Data Augmentation

Abstract:
Large multimodal models (LMMs) often struggle to recognize novel concepts, as they rely on pretrained knowledge and have limited ability to capture subtle visual details. Domain-specific knowledge gaps in training also make them prone to confusing visually similar, commonly misrepresented, or low-resource concepts. To help LMMs better align nuanced visual features with language, improving their ability to recognize and reason about novel or rare concepts, we propose a Contrastive visual Data Augmentation (CoDA) strategy. CoDA extracts key contrastive textual and visual features of target concepts against the known concepts they are misrecognized as, and then uses multimodal generative models to produce targeted synthetic data. Automatic filtering of extracted features and augmented images is implemented to guarantee their quality, as verified by human annotators. We show the effectiveness and efficiency of CoDA on low-resource concept and diverse scene recognition datasets including INaturalist and SUN. We additionally collect NovelSpecies, a benchmark dataset consisting of newly discovered animal species that are guaranteed to be unseen by LMMs. LLaVA-1.6 1-shot updating results on these three datasets show CoDA significantly improves SOTA visual data augmentation strategies by 12.3% (NovelSpecies), 5.1% (SUN), and 6.0% (iNat) absolute gains in accuracy.

Paperid:206

Authors:Weizhi Gao · Zhichao Hou · Junqi Yin · Feiyi Wang · Linyu Peng · Xiaorui Liu

Title: Modulated Diffusion: Accelerating Generative Modeling with Modulated Quantization

Abstract:
Diffusion models have emerged as powerful generative models, but their high computation cost in iterative sampling remains a significant bottleneck. In this work, we present an indepth and insightful study of state-of-the-art acceleration techniques for diffusion models, including caching and quantization, revealing their limitations in computation error and generation quality. To break these limits, this work introduces Modulated Diffusion (MoDiff), an innovative, rigorous, and principled framework that accelerates generative modeling through modulated quantization and error compensation. MoDiff not only inherents the advantages of existing caching and quantization methods but also serves as a general framework to accelerate all diffusion models. The advantages of MoDiff are supported by solid theoretical insight and analysis. In addition, extensive experiments on CIFAR-10 and LSUN demonstrate that MoDiff significant reduces activation quantization from 8 bits to 3 bits without performance degradation in post-training quantization (PTQ). Our implementation is available at https://anonymous.4open.science/r/MoDiff-78DF/ and will be make publicly available.

Paperid:207

Authors:Chengying Fang · Wenke Huang · Guancheng Wan · Yihao Yang · Mang Ye

Title: FedPHA: Federated Prompt Learning for Heterogeneous Client Adaptation

Abstract:
Federated Prompt Learning (FPL) adapts pretrained Vision-Language Models (VLMs) to federated learning through prompt tuning, leveraging their transferable representations and strong generalization capabilities. Traditional methods often require uniform prompt lengths for federated aggregation, limiting adaptability to clients with diverse prompt lengths and distribution biases. In this paper, we proposeFederatedPrompt Learning forHeterogeneous ClientAdaptation (FedPHA), a novel framework that combines a fixed-length global prompt for efficient aggregation with local prompts of varying lengths to capture client-specific data characteristics. Additionally, FedPHA designs Singular Value Decomposition (SVD) based projection and bidirectional alignment to disentangle global conflicts arising from client heterogeneity, ensuring that personalized client tasks effectively utilize non-harmful global knowledge. This approach ensures that global knowledge improves model generalization while local knowledge preserves local optimization. Experimental results validate the effectiveness of FedPHA in achieving a balance between global and personalized knowledge in federated learning scenarios.

Paperid:208

Authors:Yixin Chen · Wenjing Chen · Alan Kuhnle

Title: Breaking Barriers: Combinatorial Algorithms for Non-monotone Submodular Maximization with Sublinear Adaptivity and $1/e$ Approximation

Abstract:
Abstract:With the rapid growth of data in modern applications, parallel combinatorial algorithms for maximizing nonmonotone submodular functions have gained significant attention. The state-of-the-art approximation ratio of $1/e$ is currently achieved only by a continuous algorithm (Ene & Nguyen, 2020) with adaptivity $\mathcal O (log(n))$. In this work, we focus on size constraints and propose a $(1/4 − \varepsilon)$-approximation algorithm with high probability for this problem, as well as the first randomized parallel combinatorial algorithm achieving a $1/e − \varepsilon$ approximation ratio, which bridges the gap between continuous and combinatorial approaches. Both algorithms achieve $\mathcal O (log(n) log(k))$ adaptivity and $\mathcal O (n log(n) log(k))$ query complexity. Empirical results show our algorithms achieve competitive objective values, with the first algorithm particularly efficient in queries.

Paperid:209

Authors:Hancheng Min · Rene Vidal

Title: Gradient Flow Provably Learns Robust Classifiers for Orthonormal GMMs

Abstract:
Abstract:Deep learningbased classifiers are known to be vulnerable to adversarial attacks. Existing methods for defending against such attacks require adding a defense mechanism or modifying the learning procedure (e.g., by adding adversarial examples). This paper shows that for certain data distributions one can learn a provably robust classifier using standard learning methods and without adding a defense mechanism. More specifically, this paper addresses the problem of finding a robust classifier for a binary classification problem in which the data comes from an isotropic mixture of Gaussians with orthonormal cluster centers. First, we characterize the largest $\ell_2$-attack any classifier can defend against while maintaining high accuracy, and show the existence of optimal robust classifiers achieving this maximum $\ell_2$-robustness. Next, we show that given data sampled from the orthonormal Gaussian mixture model, gradient flow on a two-layer network with a polynomial ReLU activation and without adversarial examples provably finds an optimal robust classifier.

Paperid:210

Authors:Guangzhi Sun · Yudong Yang · Jimin Zhuang · Changli Tang · Yixuan Li · Wei Li · Zejun MA · Chao Zhang

Title: video-SALMONN-o1: Reasoning-enhanced Audio-visual Large Language Model

Abstract:
While recent advancements in reasoning optimization have significantly enhanced the capabilities of large language models (LLMs), existing efforts to improve reasoning have been limited to solving mathematical problems and focusing on visual graphical inputs, neglecting broader applications in general video understanding. This paper proposes videoSALMONN-o1, the first open-source reasoning-enhanced audio-visual LLM designed for general video understanding tasks. To enhance its reasoning abilities, we develop a reasoning-intensive dataset featuring challenging audio-visual questions with step-by-step solutions. We also propose process direct preference optimization (pDPO), which leverages contrastive step selection to achieve efficient step-level reward modelling tailored for multimodal inputs. Additionally, we introduce RivaBench, the first reasoning-intensive video understanding benchmark, featuring over 4,000 high-quality, expert-curated question-answer pairs across scenarios such as standup comedy, academic presentations, and synthetic video detection. video-SALMONN-o1 achieves 3-8% accuracy improvements over the LLaVA-OneVision baseline across different video reasoning benchmarks. Besides, pDPO achieves 6-8% improvements compared to the supervised fine-tuning model on RivaBench. Enhanced reasoning enables video-SALMONN-o1 zero-shot synthetic video detection capabilities.

Paperid:211

Authors:Antoine Wehenkel · Juan L. Gamella · Ozan Sener · Jens Behrmann · Guillermo Sapiro · Jörn Jacobsen · Marco Cuturi

Title: Addressing Misspecification in Simulation-based Inference through Data-driven Calibration

Abstract:
Driven by steady progress in deep generative modeling, simulationbased inference (SBI) has emerged as the workhorse for inferring the parameters of stochastic simulators. However, recent work has demonstrated that model misspecification can harm SBI's reliability, preventing its adoption in important applications where only misspecified simulators are available.This work introduces robust posterior estimation (RoPE), a framework that overcomes model misspecification with a small real-world calibration set of ground truth parameter measurements.We formalize the misspecification gap as the solution of an optimal transport (OT) problem between learned representations of real-world and simulated observations, allowing RoPE to learn a model of the misspecification without placing additional assumptions on its nature. RoPE shows how the calibration set and OT together offer a controllable balance between calibrated uncertainty and informative inference even under severely misspecified simulators. Results on four synthetic tasks and two real-world problems with ground-truth labels demonstrate that RoPE outperforms baselines and consistently returns informative and calibrated credible intervals.

Paperid:212

Authors:Yukang Yang · Declan Campbell · Kaixuan Huang · Mengdi Wang · Jonathan Cohen · Taylor Webb

Title: Emergent Symbolic Mechanisms Support Abstract Reasoning in Large Language Models

Abstract:
Many recent studies have found evidence for emergent reasoning capabilities in large language models, but debate persists concerning the robustness of these capabilities, and the extent to which they depend on structured reasoning mechanisms. To shed light on these issues, we perform a comprehensive study of the internal mechanisms that support abstract rule induction in an opensource language model (Llama3-70B). We identify an emergent symbolic architecture that implements abstract reasoning via a series of three computations. In early layers,abstraction headsconvert input tokens to abstract variables based on the relations between those tokens. In intermediate layers,symbolic induction headsperform sequence induction over these variables. Finally, in later layers,retrieval headspredict the next token by retrieving the value associated with the predicted variable. These results point toward a resolution of the longstanding debate between symbolic and neural network approaches, suggesting that emergent reasoning in neural networks depends on the emergence of symbolic mechanisms.

Paperid:213

Authors:Wenhao Shen · Wanqi Yin · Xiaofeng Yang · Cheng Chen · Chaoyue Song · Zhongang Cai · Lei Yang · Hao Wang · Guosheng Lin

Title: ADHMR: Aligning Diffusion-based Human Mesh Recovery via Direct Preference Optimization

Abstract:
Human mesh recovery (HMR) from a single image is inherently illposed due to depth ambiguity and occlusions. Probabilistic methods have tried to solve this by generating numerous plausible 3D human mesh predictions, but they often exhibit misalignment with 2D image observations and weak robustness to in-the-wild images. To address these issues, we propose ADHMR, a framework that Aligns Diffusion-based HMR model in a preference optimization manner. First, we train a human mesh prediction assessment model, HMR-Scorer, capable of evaluating predictions even for in-the-wild images without 3D annotations. We then use HMR-Scorer to create a preference dataset, where each input image has a pair of winner and loser mesh predictions. This dataset is used to finetune the base model using direct preference optimization. Moreover, HMR-Scorer also helps improve existing HMR models by data cleaning, even with fewer training samples. Extensive experiments show that ADHMR outperforms current state-of-the-art methods.

Paperid:214

Authors:Antonis Vasileiou · Ben Finkelshtein · Floris Geerts · Ron Levie · Christopher Morris

Title: Covered Forest: Fine-grained generalization analysis of graph neural networks

Abstract:
Abstract:The expressive power of messagepassing graph neural networks (MPNNs) is reasonably well understood, primarily through combinatorial techniques from graph isomorphism testing. However, MPNNs' generalization abilities---making meaningful predictions beyond the training set---remain less explored. Current generalization analyses often overlook graph structure, limit the focus to specific aggregation functions, and assume the impractical, hard-to-optimize $0$-$1$ loss function. Here, we extend recent advances in graph similarity theory to assess the influence of graph structure, aggregation, and loss functions on MPNNs' generalization abilities. Our empirical study supports our theoretical insights, improving our understanding of MPNNs' generalization properties.

Paperid:215

Authors:Anant Khandelwal

Title: FlexiClip: Locality-Preserving Free-Form Character Animation

Abstract:
Animating clipart images with seamless motion while maintaining visual fidelity and temporal coherence presents significant challenges. Existing methods, such as AniClipart, effectively model spatial deformations but often fail to ensure smooth temporal transitions, resulting in artifacts like abrupt motions and geometric distortions. Similarly, textto-video (T2V) and image-to-video (I2V) models struggle to handle clipart due to the mismatch in statistical properties between natural video and clipart styles. This paper introduces FlexiClip, a novel approach designed to overcome these limitations by addressing the intertwined challenges of temporal consistency and geometric integrity. FlexiClip extends traditional Bézier curve-based trajectory modeling with key innovations: temporal Jacobians to correct motion dynamics incrementally, continuous-time modeling via probability flow ODEs (pfODEs) to mitigate temporal noise, and a flow matching loss inspired by GFlowNet principles to optimize smooth motion transitions. These enhancements ensure coherent animations across complex scenarios involving rapid movements and non-rigid deformations. Extensive experiments validate the effectiveness of FlexiClip in generating animations that are not only smooth and natural but also structurally consistent across diverse clipart types, including humans and animals. By integrating spatial and temporal modeling with pre-trained video diffusion models, FlexiClip sets a new standard for high-quality clipart animation, offering robust performance across a wide range of visual content. Project Page: https://creative-gen.github.io/flexiclip.github.io/

Paperid:216

Authors:Geigh Zollicoffer · Kenneth Eaton · Jonathan Balloch · Julia Kim · Wei Zhou · Robert Wright · Mark Riedl

Title: Novelty Detection in Reinforcement Learning with World Models

Abstract:
Reinforcement learning (RL) using world models has found significant recent successes.However, when a sudden change to world mechanics or properties occurs then agent performance and reliability can dramatically decline.We refer to the sudden change in visual properties or state transitions as novelties.Implementing novelty detection within generated world model frameworks is a crucialtask for protecting the agent when deployed. In this paper, we propose straightforward bounding approaches to incorporate novelty detection into world model RL agents by utilizing the misalignment of the world model's hallucinated states and the true observed states as a novelty score. We provideeffective approaches to detecting novelties in a distribution of transitions learned by an agent ina world model. Finally, we show the advantage ofour work in a novel environment compared to traditional machine learning novelty detection methods as well as currently accepted RLfocused novelty detection algorithms.

Paperid:217

Authors:Aro Lee · Ji Oon Lee

Title: Fluctuations of the largest eigenvalues of transformed spiked Wigner matrices

Abstract:
We consider a spiked random matrix model obtained by applying a function entrywise to a signalplus-noise symmetric data matrix. We prove that the largest eigenvalue of this model, which we call a transformed spiked Wigner matrix, exhibits Baik-Ben Arous-Péché (BBP) type phase transition. We show that the law of the fluctuation converges to the Gaussian distribution when the effective signal-to-noise ratio (SNR) is above the critical number, and to the GOE Tracy-Widom distribution when the effective SNR is below the critical number. We provide precise formulas for the limiting distributions and also concentration estimates for the largest eigenvalues, both in the supercritical and the subcritical regimes.

Paperid:218

Authors:William Chen · Jinchuan Tian · Yifan Peng · Brian Yan · Chao-Han Yang · Shinji Watanabe

Title: OWLS: Scaling Laws for Multilingual Speech Recognition and Translation Models

Abstract:
Neural scaling laws offer valuable insights for designing robust sequence processing architectures. While these laws have been extensively characterized in other modalities, their behavior in speech remains comparatively underexplored. In this work, we introduce OWLS, an openaccess, reproducible suite of multilingual speech recognition and translation models spanning 0.25B to 18B parameters, with the 18B version being the largest speech model, to the best of our knowledge. OWLS leverages up to 360K hours of public speech data across 150 languages, enabling a systematic investigation into how data, model, and compute scaling each influence performance in multilingual speech tasks. We use OWLS to derive neural scaling laws, showing how final performance can be reliably predicted when scaling. One of our key findings is that scaling enhances performance on low-resource languages/dialects, helping to mitigate bias and improve the accessibility of speech technologies. Finally, we show how OWLS can be used to power new research directions by discovering emergent abilities in large-scale speech models.

Paperid:219

Authors:Youhe Jiang · Fangcheng Fu · Xiaozhe Yao · Guoliang HE · Xupeng Miao · Ana Klimovic · Bin Cui · Binhang Yuan · Eiko Yoneki

Title: Demystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs

Abstract:
Recent advancements in Large Language Models (LLMs) have led to increasingly diverse requests, accompanied with varying resource (compute and memory) demands to serve them. However, this in turn degrades the costefficiency of LLM serving as common practices primarily rely on homogeneous GPU resources. In response to this problem, this work conducts a thorough study about serving LLMs over heterogeneous GPU resources on cloud platforms. The rationale is that different GPU types exhibit distinct compute and memory characteristics, aligning well with the divergent resource demands of diverse requests. Particularly, through comprehensive benchmarking, we discover that the cost-efficiency of LLM serving can be substantially optimized by meticulously determining GPU composition, deployment configurations, and workload assignments. Subsequently, we design a scheduling algorithm via mixed-integer linear programming, aiming at deducing the most cost-efficient serving plan under the constraints of price budget and real-time GPU availability. Remarkably, our approach effectively outperforms homogeneous and heterogeneous baselines under a wide array of scenarios, covering diverse workload traces, varying GPU availablilities, and multi-model serving. This casts new light on more accessible and efficient LLM serving over heterogeneous cloud resources.

Paperid:220

Authors:Zheng Lian · Haoyu Chen · Lan Chen · Haiyang Sun · Licai Sun · Yong Ren · Zebang Cheng · Bin Liu · Rui Liu · Xiaojiang Peng · Jiangyan Yi · Jianhua Tao

Title: AffectGPT: A New Dataset, Model, and Benchmark for Emotion Understanding with Multimodal Large Language Models

Abstract:
The emergence of multimodal large language models (MLLMs) advances multimodal emotion recognition (MER) to the next level—from naive discriminative tasks to complex emotion understanding with advanced video understanding abilities and natural language description. However, the current community suffers from a lack of largescale datasets with intensive, descriptive emotion annotations, as well as a multimodal-centric framework to maximize the potential of MLLMs for emotion understanding. To address this, we establish a new benchmark for MLLM-based emotion understanding with a novel dataset (MER-Caption), and a new model (AffectGPT). Utilizing our model-based crowd-sourcing data collection strategy, we construct the largest descriptive emotion dataset to date (by far), featuring over 2K fine-grained emotion categories across 115K samples. We also introduce the AffectGPT model, designed with pre-fusion operations to enhance multimodal integration. Finally, we present MER-UniBench, a unified benchmark with evaluation metrics tailored for both typical MER tasks and the free-form, natural language output style of MLLMs. Extensive experimental results demonstrate AffectGPT's robust performance across various MER tasks. We are publicly releasing both the AffectGPT model and the MER-Caption dataset to foster further research and development in emotion understanding.

Paperid:221

Authors:Honghua Dong · Jiacheng Yang · Xun Deng · Yuhe Jiang · Gennady Pekhimenko · Fan Long · Xujie Si

Title: TypyBench: Evaluating LLM Type Inference for Untyped Python Repositories

Abstract:
Type inference for dynamic languages like Python is a persistent challenge in software engineering. While large language models (LLMs) have shown promise in code understanding, their type inference capabilities remain underexplored. We introduceTypyBench, a benchmark designed to evaluate LLMs' type inference across entire Python repositories.TypyBenchfeatures two novel metrics:TypeSim, which captures nuanced semantic relationships between predicted and ground truth types, andTypeCheck, which assesses type consistency across codebases. Our evaluation of various LLMs on a curated dataset of 50 highquality Python repositories reveals that, although LLMs achieve decentTypeSimscores, they struggle with complex nested types and exhibit significant type consistency errors. These findings suggest that future research should shift focus from improving type similarity to addressing repository-level consistency.TypyBenchprovides a foundation for this new direction, offering insights into model performance across different type complexities and usage contexts.

Paperid:222

Authors:Max Ruiz Luyten · Antonin Berthon · M van der Schaar

Title: Strategic Planning: A Top-Down Approach to Option Generation

Abstract:
Realworld human decision-making often relies on strategic planning, wherehigh-levelgoals guide the formulation of sub-goals and subsequent actions, as evidenced by domains such as healthcare, business, and urban policy. Despite notable successes in controlled settings, conventional reinforcement learning (RL) follows abottom-upparadigm, which can struggle to adapt to real-world complexities such as sparse rewards and limited exploration budgets. While methods like hierarchical RL and environment shaping provide partial solutions, they frequently rely on either ad-hoc designs (e.g. choose the set of high-level actions) or purely data-driven discovery of high-level actions that still requires significant exploration. In this paper, we introduce aTop-Downframework for RL that explicitly leverageshuman-like strategyto reduce sample complexity, guide exploration, and enable high-level decision-making. We first formalize theStrategy Problem, which frames policy generation as finding distributions over policies that balancespecificityandvalue. Building on this definition, we propose theStrategistagent—an iterative framework that leverages large language models (LLMs) to synthesize domain knowledge into a structured representation of actionable strategies and sub-goals. We further develop areward shaping methodologythat translates these strategies expressed in natural language into quantitative feedback for RL methods. Empirically, we demonstrate a significantly faster convergence than conventional PPO. Taken together, our findings highlight thattop-down strategic explorationopens new avenues for enhancing RL on real-world decision problems.

Paperid:223

Authors:Jerry Yao-Chieh Hu · Bo-Yu Chen · Dennis Wu · Feng Ruan · Han Liu

Title: Nonparametric Modern Hopfield Models

Abstract:
Abstract:We present a nonparametric interpretation for deep learning compatible modern Hopfield models and utilize this new perspective to debut efficient variants. Our key contribution stems from interpreting the memory storage and retrieval processes in modern Hopfield models as a nonparametric regression problem subject to a set of querymemory pairs.Interestingly,our framework not only recovers the known results from the original dense modern Hopfield model but also fills the void in the literature regarding efficient modern Hopfield models, by introducing \textit{sparse-structured} modern Hopfield models with sub-quadratic complexity.We establish that this sparse model inherits the appealing theoretical properties of its dense analogue --- connection with transformer attention, fixed point convergence and exponential memory capacity.Additionally, we showcase the versatility of our framework by constructing a family of modern Hopfield models as extensions, including linear, random masked, top-$K$ and positive random feature modern Hopfield models.Empirically, we validate our framework in both synthetic and realistic settings for memory retrieval and learning tasks.

Paperid:224

Authors:Tao Feng · Yexin Wu · Guanyu Lin · Jiaxuan You

Title: Graph World Model

Abstract:
World models (WMs) demonstrate strong capabilities in prediction, generation, and planning tasks.Existing WMs primarily focus on unstructured data while cannot leverage the ubiquitous structured data, often represented as graphs, in the digital world. While multiple graph foundation models have been proposed, they focus on graph learning tasks and cannot extend to diverse multimodal data and interdisciplinary tasks. To address these challenges, we propose the Graph World Model (GWM), a world model that supports both unstructured and graph-structured states with multi-modal information and represents diverse tasks as actions. The core of a GWM is a generic message-passing algorithm to aggregate structured information, either over a unified multi-modal token space by converting multi-modal data into text (GWM-T) or a unified multi-modal embedding space by modality-specific encoders (GWM-E). Notably, GWM introduces action nodes to support diverse tasks, where action nodes are linked to other nodes via direct reference or similarity computation. Extensive experiments on 6 tasks from diverse domains, including multi-modal generation and matching, recommendation, graph prediction, multi-agent, retrieval-augmented generation, and planning and optimization, show that the same GWM outperforms or matches domain-specific baselines' performance, benefits from multi-hop structures, and demonstrate strong zero-shot/few-shot capabilities on unseen new tasks.

Paperid:225

Authors:Anle Ke · Xu Zhang · Tong Chen · Ming Lu · Chao Zhou · Jiawen Gu · Zhan Ma

Title: Ultra Lowrate Image Compression with Semantic Residual Coding and Compression-aware Diffusion

Abstract:
Existing multimodal large modelbased image compression frameworks often rely on a fragmented integration of semantic retrieval, latent compression, and generative models, resulting in suboptimal performance in both reconstruction fidelity and coding efficiency. To address these challenges, we propose a residual-guided ultra lowrate image compression named ResULIC, which incorporates residual signals into both semantic retrieval and the diffusion-based generation process. Specifically, we introduce Semantic Residual Coding (SRC) to capture the semantic disparity between the original image and its compressed latent representation. A perceptual fidelity optimizer is further applied for superior reconstruction quality. Additionally, we present the Compression-aware Diffusion Model (CDM), which establishes an optimal alignment between bitrates and diffusion time steps, improving compression-reconstruction synergy. Extensive experiments demonstrate the effectiveness of ResULIC, achieving superior objective and subjective performance compared to state-of-the-art diffusion-based methods with -80.7\%, -66.3\% BD-rate saving in terms of LPIPS and FID.

Paperid:226

Authors:Jiachen Guo · Xiaoyu Xie · Chanwook Park · Hantao Zhang · Matthew Politis · Gino Domel · Jiachen Guo

Title: Interpolating Neural Network-Tensor Decomposition (INN-TD): a scalable and interpretable approach for large-scale physics-based problems

Abstract:
Deep learning has been extensively employed as a powerful function approximator for modeling physicsbased problems described by partial differential equations (PDEs). Despite their popularity, standard deep learning models often demand prohibitively large computational resources and yield limited accuracy when scaling to large-scale, high-dimensional physical problems. Their black-box nature further hinders their application in industrial problems where interpretability and high precision are critical. To overcome these challenges, this paper introduces Interpolation Neural Network-Tensor Decomposition (INN-TD), a scalable and interpretable framework that has the merits of both machine learning and finite element methods for modeling large-scale physical systems. By integrating locally supported interpolation functions from finite element into the network architecture, INN-TD achieves a sparse learning structure with enhanced accuracy, faster training/solving speed, and reduced memory footprint. This makes it particularly effective for tackling large-scale high-dimensional parametric PDEs in training, solving, and inverse optimization tasks in physical problems where high precision is required.

Paperid:227

Authors:Longlin Yu · Jiajun Zha · Tong Yang · Tianyu Xie · Xiangyu Zhang · Gary Chan · Cheng Zhang

Title: Continuous Semi-Implicit Models: A Path Towards Consistency

Abstract:
Semiimplicit distributions have shown great promise in variational inference and generative modeling. Hierarchical semi-implicit models, which stack multiple semi-implicit layers, enhance the expressiveness of semi-implicit distributions and can be used to accelerate diffusion models given pretrained score networks. However, their sequential training often suffers from slow convergence. In this paper, we introduce CoSIM, a continuous semi-implicit model that extends hierarchical semi-implicit models into a continuous framework. By incorporating a continuous transition kernel, CoSIM enables efficient, simulation-free training. Furthermore, we show that CoSIM achieves consistency with a carefully designed transition kernel, introducing a novel training scheme for consistency models. Extensive experiments on image generation demonstrate that CoSIM performs on par or better than existing diffusion model acceleration methods, achieving superior performance on FD-DINOv2.

Paperid:228

Authors:Samuel Miserendino · Michele Wang · Tejal Patwardhan · Johannes Heidecke

Title: SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?

Abstract:
Abstract:We introduce SWELancer, a benchmark of over 1400 freelance software engineering tasks from Upwork, valued at \\\$1 million USD total in real-world payouts. SWE-Lancer encompasses both independent engineering tasks — ranging from \\\$50 bug fixes to \\\$32000 feature implementations — and managerial tasks, where models choose between technical implementation proposals. Independent tasks are graded with end-to-end tests triple-verified by experienced software engineers, while managerial decisions are assessed against the choices of the original hired engineering managers. We evaluate model performance and find that frontier models are still unable to solve the majority of tasks. To facilitate future research, we open-source a unified Docker image and a public evaluation split. By mapping model performance to monetary value, we hope SWE-Lancer enables greater research into the economic impact of AI model development.

Paperid:229

Authors:Marco Cipriano · Moritz Feuerpfeil · Gerard de Melo

Title: Vector Grimoire: Codebook-based Shape Generation under Raster Image Supervision

Abstract:
Scalable Vector Graphics (SVG) is a popular format on the web and in the design industry. However, despite the great strides made in generative modeling, SVG has remained underexplored due to the discrete and complex nature of such data. We introduce GRIMOIRE, a textguided SVG generative model that is comprised of two modules: A Visual Shape Quantizer (VSQ) learns to map raster images onto a discrete codebook by reconstructing them as vector shapes, and an Auto-Regressive Transformer (ART) models the joint probability distribution over shape tokens, positions and textual descriptions, allowing us to generate vector graphics from natural language. Unlike existing models that require direct supervision from SVG data, GRIMOIRE learns shape image patches using only raster image supervision which opens up vector generative modeling to significantly more data. We demonstrate the effectiveness of our method by fitting GRIMOIRE for closed filled shapes on the MNIST and Emoji, and for outline strokes on icon and font data, surpassing previous image-supervised methods in generative quality and vector-supervised approach in flexibility.

Paperid:230

Authors:Hai Ci · Yiren Song · Pei Yang · Jinheng Xie · Mike Zheng Shou

Title: WMAdapter: Adding WaterMark Control to Latent Diffusion Models

Abstract:
Watermarking is essential for protecting the copyright of AIgenerated images. We propose WMAdapter, a diffusion model watermark plugin that embeds user-specified watermark information seamlessly during the diffusion generation process. Unlike previous methods that modify diffusion modules to incorporate watermarks, WMAdapter is designed to keep all diffusion components intact, resulting in sharp, artifact-free images. To achieve this, we introduce two key innovations: (1) We develop a contextual adapter that conditions on the content of the cover image to generate adaptive watermark embeddings. (2) We implement an additional finetuning step and a hybrid finetuning strategy that suppresses noticeable artifacts while preserving the integrity of the diffusion components. Empirical results show that WMAdapter provides strong flexibility, superior image quality, and competitive watermark robustness.

Paperid:231

Authors:Jing Xu · Jiazheng Li · Jingzhao Zhang

Title: Scalable Model Merging with Progressive Layer-wise Distillation

Abstract:
Model merging offers an effective way to integrate the capabilities of multiple finetuned models. However, the performance degradation of the merged model remains a challenge, particularly when none or few data are available. This paper first highlights the necessity of domain-specific data for model merging by proving that data-agnostic algorithms can have arbitrarily bad worst-case performances. Building on this theoretical insight, we explore the relationship between model merging and distillation, introducing a novel few-shot merging algorithm, ProDistill (Progressive Layer-wise Distillation). Unlike common belief that layer-wise training hurts performance, we show that layer-wise teacher-student distillation not only enhances the scalability but also improves model merging performance. We conduct extensive experiments to show that compared to existing few-shot merging methods,ProDistill achieves state-of-the-art performance, with up to 6.14\% and 6.61\% improvements in vision and NLU tasks. Furthermore, we extend the experiments to models with over 10B parameters, showcasing the exceptional scalability of ProDistill.

Paperid:232

Authors:Sreyan Ghosh · Zhifeng Kong · Sonal Kumar · S Sakshi · Jaehyeon Kim · Wei Ping · Rafael Valle · Dinesh Manocha · Bryan Catanzaro

Title: Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities

Abstract:
Understanding and reasoning over nonspeech sounds and music are crucial for both humans and AI agents to interact effectively with their environments. In this paper, we introduce Audio Flamingo 2 (AF2), an Audio-Language Model (ALM) with advanced audio understanding and reasoning capabilities. AF2 leverages (i) a custom CLAP model, (ii) synthetic AQA data for fine-grained audio reasoning, and (iii) a multi-stage curriculum learning strategy. AF2 achieves state-of-the-art performance with only a 3B parameter small language model, surpassing large open-source and proprietary models across 20+ benchmarks. Next, for the first time, we extend audio understanding to long audio segments (30 secs - 5 mins) and propose LongAudio, a large and novel dataset for training ALMs on long audio captioning and question-answering tasks. Fine-tuning AF2 on LongAudio leads to exceptional performance on our proposed LongAudioBench, an expert annotated benchmark for evaluating ALMs on long audio understanding capabilities. We conduct extensive ablation studies to confirm the efficacy of our approach. All code and data will be open-sourced. Demo: https://sites.google.com/view/audioflamingo2

Paperid:233

Authors:Yeyun Chen

Title: Relational Invariant Learning for Robust Solvation Free Energy Prediction

Abstract:
Predicting the solvation free energy of molecules using graph neural networks holds significant potential for advancing drug discovery and the design of novel materials. While previous models have demonstrated success on independent and identically distributed (IID) datasets, their performance in outof-distribution (OOD) scenarios remains largely unexplored. We propose a novel Relational Invariant Learning framework (RILOOD) to enhance OOD generalization in solvation free energy prediction. RILOOD comprises three key components: (i) a mixup-based conditional modeling module that integrates diverse environments, (ii) a novel multi-granularity refinement strategy that extends beyond core substructures to enable context-aware representation learning for capturing multi-level interactions, and (iii) an invariant learning module that identifies robust patterns generalizable to unseen environments. Extensive experiments demonstrate that RILOOD significantly outperforms state-of-the-art methods across various distribution shifts, highlighting its effectiveness in improving solvation free energy prediction under diverse conditions.

Paperid:234

Authors:Quoc Tran-Dinh

Title: Variance-Reduced Forward-Reflected-Backward Splitting Methods for Nonmonotone Generalized Equations

Abstract:
Abstract:We develop two novel stochastic variancereduction methods to approximate solutions of a class of nonmonotone [generalized] equations. Our algorithms leverage a new combination of ideas from the forward-reflected-backward splitting method and a class of unbiased variance-reduced estimators. We construct two new stochastic estimators within this class, inspired by the well-known SVRG and SAGA estimators. These estimators significantly differ from existing approaches used in minimax and variational inequality problems. By appropriately choosing parameters, both algorithms achieve state-of-the-art oracle complexity of $\mathcal{O}(n + n^{2/3} \epsilon^{-2})$ for obtaining an $\epsilon$-solution in terms of the operator residual norm for a class of nonmonotone problems, where $n$ is the number of summands and $\epsilon$ signifies the desired accuracy. This complexity aligns with the best-known results in SVRG and SAGA methods for stochastic nonconvex optimization. We test our algorithms on two numerical examples and compare them with existing methods. The results demonstrate promising improvements offered by the new methods compared to their competitors.

Paperid:235

Authors:Runxi Cheng · Feng Xiong · Yongxian Wei · Wanyun Zhu · Chun Yuan

Title: Whoever Started the interference Should End It: Guiding Data-Free Model Merging via Task Vectors

Abstract:
Model merging seeks to integrate taskspecific expert models into a unified architecture while preserving multi-task generalization capabilities, yet parameter interference between constituent models frequently induces performance degradation. Although prior work has explored many merging strategies, resolving interference without additional data for retraining or test-time computation remains challenging. In this paper, we theoretically demonstrate that the task vectors of the linear layer constitute an approximate linear subspace for its corresponding input. Therefore, we can minimize interference under the guidance of task vectors. Based on this insight, we propose \textbf{WUDI-Merging} (\textbf{W}hoever started the interference sho\textbf{U}ld en\textbf{D} \textbf{I}t), a simple yet effective model merging method that eliminates interference without any additional data or rescaling coefficients. Comprehensive empirical evaluations across vision and language benchmarks demonstrate our method's superiority, achieving state-of-the-art performance in data-free model merging scenarios (average 10.9\% improvement versus baseline methods) while even outperforming mainstream test-time adaptation approaches by 3.3\%, and only very few computing resources are required. The code will be publicly available soon.

Paperid:236

Authors:Naoki Nishikawa · Yujin Song · Kazusato Oko · Denny Wu · Taiji Suzuki

Title: Nonlinear transformers can perform inference-time feature learning: a case study of in-context learning on single-index models

Abstract:
Abstract:Pretrained transformers have demonstrated the ability to implement various algorithms at inference time without parameter updates. While theoretical works have established this capability through constructions and approximation guarantees, the optimization and statistical efficiency aspects remain understudied. In this work, we investigate how transformers learn features incontext -- a key mechanism underlying their inference-time adaptivity. We focus on the in-context learning of single-index models $y=\sigma_*(\langle\boldsymbol x,\boldsymbol\beta\rangle)$, which are low-dimensional nonlinear functions parameterized by feature vector $\boldsymbol\beta$. We prove that transformers pretrained by gradient-based optimization can perform *inference-time feature learning*, i.e., extract information of $\boldsymbol\beta$ solely from test prompts (despite $\boldsymbol\beta$ varying across different prompts), hence achieving inference-time statistical efficiency that surpasses any non-adaptive (fixed-basis) algorithms such as kernel methods. Moreover, we show that the inference-time sample complexity surpasses the Correlational Statistical Query (CSQ) lower bound, owing to nonlinear label transformations naturally induced by the self-attention mechanism.

Paperid:237

Authors:Jiujia Zhang · Ashok Cutkosky

Title: Unconstrained Robust Online Convex Optimization

Abstract:
Abstract:This paper addresses online learning with ''corrupted'' feedback. Our learner is provided with potentially corrupted gradients $\tilde g_t$ instead of the ''true'' gradients $g_t$. We make no assumptions about how the corruptions arise: they could be the result of outliers, mislabeled data, or even malicious interference. We focus on the difficult ``unconstrained'' setting in which our algorithm must maintain low regret with respect to any comparison point $\|u\| \in \R^d$. The unconstrained setting is significantly more challenging as existing algorithms suffer extremely high regret even with very tiny amounts of corruption (which is not true in the case of a bounded domain). Our algorithms guarantee regret $ \|u\|G (\sqrt{T} + k) $ when $G \ge \max_t \|g_t\|$ is known, where $k$ is a measure of the total amount of corruption. When $G$ is unknown we incur an extra additive penalty of $(\|u\|^2+G^2) k$.

Paperid:238

Authors:Xudong Gong · Sen Yang · Feng Dawei · Kele Xu · Bo Ding · Huaimin Wang · Yong Dou

Title: Improving the Continuity of GoalAchievement Ability via Policy Self-Regularization for Goal-Conditioned Reinforcement Learning

Abstract:
This paper addresses the challenge of discontinuity in goalachievement capabilities observed in Goal-conditioned Reinforcement Learning (GCRL) algorithms. Through a theoretical analysis, we identify that the reuse of successful trajectories or policies during training can aid in achieving adjacent goals of achievable goals. However, the policy discrepancy between achievable and adjacent goals must be carefully managed to avoid both overly trivial and excessively large differences, which can respectively hinder policy performance. To tackle this issue, we propose a margin-based policy self-regularization approach that optimizes the policy discrepancies between adjacent desired goals to a minimal acceptable threshold. This method can be integrated into popular GCRL algorithms, such as GC-SAC, HER, and GC-PPO. Systematic evaluations across two robotic arm control tasks and a complex fixed-wing aircraft control task demonstrate that our approach significantly improves the continuity of goal-achievement abilities of GCRL algorithms, thereby enhancing their overall performance.

Paperid:239

Authors:Paolo Pellizzoni · Simone Moretti · Francesco Silvestri

Title: Dimensionality Reduction on Complex Vector Spaces for Euclidean Distance with Dynamic Weights

Abstract:
Abstract:The weighted Euclidean norm $\lVert x\rVert_w$ of a vector $x\in \mathbb{R}^d$ with weights $w\in \mathbb{R}^d$ is the Euclidean norm where the contribution of each dimension is scaled by a given weight. Approaches to dimensionality reduction that satisfy the Johnson–Lindenstrauss (JL) lemma can be easily adapted to the weighted Euclidean distance if weights are known and fixed: it suffices to scale each dimension of the input vectors according to the weights, and then apply any standard approach.However, this is not the case when weights are unknown during the dimensionality reduction or might dynamically change.In this paper, we address this issue by providing an approach that maps vectors into a smaller complex vector space and allows to retrieve a JLlike estimate for the weighted Euclidean distance once weights are revealed.Our result are based on the decomposition of the complex dimensionality reduction into several Rademacher chaos random variables, which are studied using novel concentration inequalities for sums of independent Rademacher chaoses.

Paperid:240

Authors:Nuoya Xiong · Aarti Singh

Title: Projection Optimization: A General Framework for Multi-Objective and Multi-Group RLHF

Abstract:
Reinforcement Learning with Human Feedback (RLHF) is a widely used finetuning approach that aligns machine learning models, particularly Language Models (LMs) with human preferences. There are typically multiple objectives driving the preference, hence humans find it easier to express per-objective comparisons rather than a global preference between two choices, e.g. compare two papers on their novelty, clarity, correctness, etc. Multi-Objective RLHF aims to use per-objective preference feedback and achieve a Pareto optimal tradeoff among these objectives by aggregating them into a single unified objective for optimization. However, nearly all prior works rely on linear aggregation, which rules out policies that favor specific objectives such as the worst one. The only existing approach using non-linear aggregation is computationally expensive due to its reward-based nature and the need for retraining whenever the aggregation parameters change. In this work, we address this limitation by transforming the non-linear aggregation maximization problem into a series of sub-problems. Each sub-problem involves only linear aggregation, making it computationally efficient to solve. We further extend our framework to handle multi-group scenarios, where each group has distinct weights for the objectives. Our method enables achieving consensus or maximizing the aggregated objective across all groups. Theoretically, we demonstrate that our algorithmic framework achieves sublinear regret and can be easily adapted to a reward-free algorithm. Empirically, leveraging our theoretical insights, we propose a nearly training-free algorithm once the optimal policies for individual objectives are obtained.

Paperid:241

Authors:Giulio Starace · Oliver Jaffe · Dane Sherburn · James Aung · Jun Shern Chan · Leon Maksin · Rachel Dias · Evan Mays · Benjamin Kinsella · Wyatt Thompson · Johannes Heidecke · Amelia Glaese · Tejal Patwardhan

Title: PaperBench: Evaluating AI’s Ability to Replicate AI Research

Abstract:
We introduce PaperBench, a benchmark for evaluating AI agents' ability to replicate stateof-the-art machine learning research. We curate 18 Spotlight and Oral papers from ICML 2024 to build a dataset of papers representing the current frontier of machine learning (ML) research. Replicating these papers requires real-world ML engineering skills such as understanding the literature, preparing datasets, training models, and running experiments. We manually createrubricswith the author(s) of each paper, which hierarchically decompose each replication task into smaller sub-tasks along with clear grading criteria. We develop an LLM-basedjudgeto grade replication attempts against rubrics, and assess the judge's performance by creating a separate benchmark for judges. We evaluate several frontier models on PaperBench, finding that the best-performing agent -- Claude Sonnet 3.5 with open-source scaffolding -- achieves a replication score of 14.1\%. Finally, we recruit ML PhDs to attempt a 5-paper subset of PaperBench, finding that human participants achieve 22.0\%, compared to Claude 3.5 Sonnet which scores 11.0\% on this subset. We open-source our code to facilitate future research in understanding the ML engineering capabilities of AI agents.

Paperid:242

Authors:Deqian Kong · Minglu Zhao · Dehong Xu · Bo Pang · Shu Wang · Edouardo Honig · Zhangzhang Si · Chuan Li · Jianwen Xie · Sirui Xie · Ying Nian Wu

Title: Scalable Language Models with Posterior Inference of Latent Thought Vectors

Abstract:
We propose a novel family of language models, LatentThought Language Models (LTMs), which incorporate explicit latent thought vectors that guide the autoregressive generation of ground tokens through a Transformer decoder. Training employs a dual-rate optimization process within the classical variational Bayes framework: fast learning of local variational parameters for the posterior distribution of latent vectors, and slow learning of global decoder parameters. Empirical studies reveal that LTMs possess additional scaling dimensions beyond traditional LLMs, yielding a structured design space. Higher sample efficiency can be achieved by increasing training compute per token, with further gains possible by trading model size for more inference steps. Designed based on these scaling properties, LTMs demonstrate superior sample and parameter efficiency compared to conventional autoregressive models and discrete diffusion models. They significantly outperform these counterparts in validation perplexity and zero-shot language modeling. Additionally, LTMs exhibit emergent few-shot in-context reasoning capabilities that scale with model and latent size, and achieve competitive performance in unconditional and conditional text generation.

Paperid:243

Authors:Viacheslav Meshchaninov · Pavel Strashnov · Andrey Shevtsov · Fedor Nikolaev · Nikita Ivanisenko · Olga Kardymon · Dmitry Vetrov

Title: Diffusion on language model encodings for protein sequence generation

Abstract:
Proteinsequencedesign has seen significant advances through discrete diffusion and autoregressive approaches, yet the potential of continuous diffusion remains underexplored. Here, we presentDiMA, a latent diffusion framework that operates on protein language model representations. Through systematic exploration of architectural choices and diffusion components, we develop a robust methodology that generalizes across multiple protein encoders ranging from 8M to 3B parameters. We demonstrate that our framework achieves consistently high performance across sequenceonly (ESM-2, ESMc), dual-decodable (CHEAP), and multimodal (SaProt) representations using the same architecture and training approach. We conduct extensive evaluation of existing methods alongsideDiMAusing multiple metrics across two protein modalities, covering quality, diversity, novelty, and distribution matching of generated proteins.DiMAconsistently produces novel, high-quality and diverse protein sequences and achieves strong results compared to baselines such as autoregressive, discrete diffusion and flow matching language models. The model demonstrates versatile functionality, supporting conditional generation tasks including protein family-generation, motif scaffolding and infilling, and fold-specific sequence design, despite being trained solely on sequence data. This work provides a universal continuous diffusion framework for protein sequence generation, offering both architectural insights and practical applicability across various protein design scenarios.

Paperid:244

Authors:Yushu Shi · Chao Huang · Jie Wen · Wei Wang · Yong Xu · Xiaochun Cao

Title: Ex-VAD: Explainable Fine-grained Video Anomaly Detection Based on Visual-Language Models

Abstract:
With advancements in visual language models (VLMs) and large language models (LLMs), video anomaly detection (VAD) has progressed beyond binary classification to finegrained categorization and multidimensional analysis. However, existing methods focus mainly on coarse-grained detection, lacking anomaly explanations. To address these challenges, we propose Ex-VAD, an Explainable Fine-grained Video Anomaly Detection approach that combines fine-grained classification with detailed explanations of anomalies. First, we use a VLM to extract frame-level captions, and an LLM converts them to video-level explanations, enhancing the model's explainability. Second, integrating textual explanations of anomalies with visual information greatly enhances the model's anomaly detection capability. Finally, we apply label-enhanced alignment to optimize feature fusion, enabling precise fine-grained detection. Experiments on the UCF-Crime and XD-Violence datasets demonstrate that Ex-VAD significantly outperforms existing methods and enhances model transparency with detailed explanatory texts. Extensive experimental results on the UCF-Crime and XD-Violence datasets demonstrate that Ex-VAD significantly outperforms existing State-of-The-Art methods.

Paperid:245

Authors:Hao Yu · Weixuan Liang · KE LIANG · Suyuan Liu · Meng Liu · Xinwang Liu

Title: On the Adversarial Robustness of Multi-Kernel Clustering

Abstract:
Multikernel clustering (MKC) has emerged as a powerful method for capturing diverse data patterns, offering robust and generalized representations of data structures. However, the increasing deployment of MKC in real-world applications raises concerns about its vulnerability to adversarial perturbations. While adversarial robustness has been extensively studied in other domains, its impact on MKC remains largely unexplored. In this paper, we address the challenge of assessing the adversarial robustness of MKC methods in a black-box setting. Specifically, we proposeAdvMKC, a novel reinforcement-learning-based adversarial attack framework designed to inject imperceptible perturbations into data and mislead MKC methods. AdvMKC leverages proximal policy optimization with an advantage function to overcome the instability of clustering results during optimization. Additionally, it introduces a generator-clusterer framework, where a generator produces adversarial perturbations, and a clusterer approximates MKC behavior, significantly reducing computational overhead. We provide theoretical insights into the impact of adversarial perturbations on MKC and validate these findings through experiments. Evaluations across seven datasets and eleven MKC methods (seven traditional and four robust) demonstrate AdvMKC's effectiveness, robustness, and transferability.

Paperid:246

Authors:Muleilan Pei · Shaoshuai Shi · Lu Zhang · Peiliang Li · Shaojie Shen

Title: GoIRL: Graph-Oriented Inverse Reinforcement Learning for Multimodal Trajectory Prediction

Abstract:
Trajectory prediction for surrounding agents is a challenging task in autonomous driving due to its inherent uncertainty and underlying multimodality. Unlike prevailing datadriven methods that primarily rely on supervised learning, in this paper, we introduce a novel Graph-oriented Inverse Reinforcement Learning (GoIRL) framework, which is an IRL-based predictor equipped with vectorized context representations. We develop a feature adaptor to effectively aggregate lane-graph features into grid space, enabling seamless integration with the maximum entropy IRL paradigm to infer the reward distribution and obtain the policy that can be sampled to induce multiple plausible plans. Furthermore, conditioned on the sampled plans, we implement a hierarchical parameterized trajectory generator with a refinement module to enhance prediction accuracy and a probability fusion strategy to boost prediction confidence. Extensive experimental results showcase our approach not only achieves state-of-the-art performance on the large-scale Argoverse & nuScenes motion forecasting benchmarks but also exhibits superior generalization abilities compared to existing supervised models.

Paperid:247

Authors:Pierre C Bellec · Takuya Koriyama

Title: Phase transitions for the existence of unregularized M-estimators in single index models

Abstract:
Abstract:This paper studies phase transitions for the existence of unregularized Mestimators under proportional asymptotics where the sample size $n$ and feature dimension $p$ grow proportionally with $n/p \to \delta \in (1, \infty)$. We study the existence of M-estimators in single-index models where the response $y_i$ depends on covariates $x_i \sim N(0, I_p)$ through an unknown index ${w} \in \mathbb{R}^p$ and an unknown link function. An explicit expression is derived for the critical threshold $\delta_\infty$ that determines the phase transition for the existence of the M-estimator, generalizing the results of Candés & Sur (2020) for binary logistic regression to other single-index models.Furthermore, we investigate the existence of a solution to the nonlinear system of equations governing the asymptotic behavior of the M-estimator when it exists. The existence of solution to this system for $\delta > \delta_\infty$ remains largely unproven outside the global null in binary logistic regression. We address this gap with a proof that the system admits a solution if and only if $\delta > \delta_\infty$, providing a comprehensive theoretical foundation for proportional asymptotic results that require as a prerequisite the existence of a solution to the system.

Paperid:248

Authors:Gábor Pituk · Vik Shirvaikar · Tom Rainforth

Title: Do Bayesian Neural Networks Actually Behave Like Bayesian Models?

Abstract:
We empirically investigate how well popular approximate inference algorithms for Bayesian Neural Networks (BNNs) respect the theoretical properties of Bayesian belief updating. We find strong evidence on synthetic regression and realworld image classification tasks that common BNN algorithms such as variational inference, Laplace approximation, SWAG, and SGLD fail to update in a consistent manner, forget about old data under sequential updates, and violate the predictive coherence properties that would be expected of Bayesian methods. These observed behaviors imply that care should be taken when treating BNNs as true Bayesian models, particularly when using them beyond static prediction settings, such as for active, continual, or transfer learning.

Paperid:249

Authors:Zehan Wang · Ziang Zhang · Tianyu Pang · Chao Du · Hengshuang Zhao · Zhou Zhao

Title: Orient Anything: Learning Robust Object Orientation Estimation from Rendering 3D Models

Abstract:
Orientation is a fundamental attribute of objects, essential for understanding their spatial pose and arrangement. However, practical solutions for estimating the orientation of openworld objects in monocular images remain underexplored. In this work, we introduce Orient Anything, the first foundation model for zero-shot object orientation estimation. A key challenge in this task is the scarcity of orientation annotations for open-world objects. To address this, we propose leveraging the vast resources of 3D models. By developing a pipeline to annotate the front face of 3D objects and render them from random viewpoints, we curate 2 million images with precise orientation annotations across a wide variety of object categories. To fully leverage the dataset, we design a robust training objective that models the 3D orientation as probability distributions over three angles and predicts the object orientation by fitting these distributions. Besides, we propose several strategies to further enhance the synthetic-to-real transfer. Our model achieves state-of-the-art orientation estimation accuracy on both rendered and real images, demonstrating impressive zero-shot capabilities across various scenarios. Furthermore, it shows great potential in enhancing high-level applications, such as understanding complex spatial concepts in images and adjusting 3D object pose.

Paperid:250

Authors:Chi Zhang · REN Lianhai · Jingpu Cheng · Qianxiao Li

Title: From Weight-Based to State-Based Fine-Tuning: Further Memory Reduction on LoRA with Parallel Control

Abstract:
The LoRA method has achieved notable success in reducing GPU memory usage by applying lowrank updates to weight matrices. Yet, one simple question remains: can we push this reduction even further? Furthermore, is it possible to achieve this while improving performance and reducing computation time? Answering these questions requires moving beyond the conventional weight-centric approach. In this paper, we present a state-based fine-tuning framework that shifts the focus from weight adaptation to optimizing forward states, with LoRA acting as a special example. Specifically, state-based tuning introduces parameterized perturbations to the states within the computational graph, allowing us to control states across an entire residual block. A key advantage of this approach is the potential to avoid storing large intermediate states in models like transformers. Empirical results across multiple architectures—including ViT, RoBERTa, LLaMA2-7B, and LLaMA3-8B—show that our method further reduces memory consumption and computation time while simultaneously improving performance. Moreover, as a result of memory reduction, we explore the feasibility to train 7B/8B models on consumer-level GPUs like Nvidia 3090, without model quantization. The code is available at an anonymous GitHub repository

Paperid:251

Authors:Niv Buchbinder · Roie Levin · Yue Yang

Title: Competitively Consistent Clustering

Abstract:
Abstract:In *fullydynamic consistent clustering*, we are given a finite metric space $(M,d)$, and a set $F\subseteq M$ of possible locations for opening centers. Data points arrive and depart, and the goal is to maintain an approximately optimal clustering solution at all times while minimizing the *recourse*, the total number of additions/deletions of centers over time. Specifically, we study fully dynamic versions of the classical $k$-center, facility location, and $k$-median problems. We design algorithms that, given a parameter $\beta\geq 1$, maintain an $O(\beta)$-approximate solution at all times, and whose total recourse is bounded by $O(\log |F| \log \Delta) \cdot OPT_{rec}^{\beta}$. Here $OPT_{rec}^{\beta}$ is the minimal recourse of an offline algorithm that maintains a $\beta$-approximate solution at all times, and $\Delta$ is the metric aspect ratio. We obtain our results via a reduction to the recently proposed *Positive Body Chasing* framework of [Bhattacharya Buchbinder Levin Saranurak, FOCS 2023], which we show gives fractional solutions to our clustering problems online. Our contribution is to round these fractional solutions while preserving the approximation and recourse guarantees. We complement our positive results with logarithmic lower bounds which show that our bounds are nearly tight.

Paperid:252

Authors:Changsheng Wang · Yihua Zhang · jinghan jia · Parikshit Ram · Dennis Wei · Yuguang Yao · Soumyadeep Pal · Nathalie Baracaldo · Sijia Liu

Title: Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning

Abstract:
Machine unlearning presents a promising approach to mitigating privacy and safety concerns in large language models (LLMs) by enabling the selective removal of targeted data or knowledge while preserving model utility. However, existing unlearning methods remain oversensitive to downstream fine-tuning, which can rapidly recover what is supposed to be unlearned information even when the fine-tuning task is entirely unrelated to the unlearning objective.To enhance robustness, we introduce the concept of `invariance' into unlearning for the first time from the perspective of invariant risk minimization (IRM), a principle for environment-agnostic training. By leveraging IRM, we develop a new invariance-regularized LLM unlearning framework, termed invariant LLM unlearning (ILU). We show that the proposed invariance regularization, even using only a single fine-tuning dataset during ILU training, can enable unlearning robustness to generalize effectively across diverse and new fine-tuning tasks at test time.A task vector analysis is also provided to further elucidate the rationale behind ILU's effectiveness. Extensive experiments on the WMDP benchmark, which focuses on removing an LLM's hazardous knowledge generation capabilities, reveal that ILU significantly outperforms state-of-the-art unlearning methods, including negative preference optimization (NPO) and representation misdirection for unlearning (RMU). Notably, ILU achieves superior unlearning robustness across diverse downstream fine-tuning scenarios (e.g., math, paraphrase detection, and sentiment analysis) while preserving the fine-tuning performance.

Paperid:253

Authors:Senyu Han · Hongchuan Zeng · Kai Yu · Lu Chen

Title: Heads up! Large Language Models Can Perform Tasks Without Your Instruction via Selective Attention Head Masking

Abstract:
Large language models (LLMs) consist of numerous Transformer modules, and while the models can perform various functions, it remains an open question of how these modules are combined to elicit distinct inherent functionalities. In this paper, we investigate the modules inside LLMs and demonstrate that, by simply masking or retaining specific attention heads during inference, LLMs can exhibit specific task functionalities without requiring explicit instructions or modifications to the model parameters. Experiments across various models and tasks reveal that LLMs inherently encode ``functional pathways'', the structured groups of interdependent attention heads that are crucial for executing specific tasks. These pathways not only govern the model's functional behaviors but also enhance parameter efficiency, as suppressing attention heads outside the pathway can improve task performance. The code is available in this repository:Anonymous Github.

Paperid:254

Authors:Ke Fan · Jinpeng Zhang · Xuefeng Zhang · Yunze Wu · Jingyu Cao · Yuan Zhou · Jianzhu Ma

Title: SafetyPolarized and Prioritized Reinforcement Learning

Abstract:
Motivated by the first priority of safety in many realworld applications, we propose \textsc{MaxSafe}, a chance-constrained bi-level optimization framework for safe reinforcement learning. \textsc{MaxSafe} first minimizes the unsafe probability and then maximizes the return among the safest policies. We provide a tailored Q-learning algorithm for the \textsc{MaxSafe} objective, featuring a novel learning process for \emph{optimal action masks} with theoretical convergence guarantees. To enable the application of our algorithm to large-scale experiments, we introduce two key techniques: \emph{safety polarization} and \emph{safety prioritized experience replay}. Safety polarization generalizes the optimal action masking by polarizing the Q-function, which assigns low values to unsafe state-action pairs, effectively discouraging their selection. In parallel, safety prioritized experience replay enhances the learning of optimal action masks by prioritizing samples based on temporal-difference (TD) errors derived from our proposed state-action reachability estimation functions. This approach efficiently addresses the challenges posed by sparse cost signals. Experiments on diverse autonomous driving and safe control tasks show that our methods achieve near-maximal safety and an optimal reward-safety trade-off.

Paperid:255

Authors:Yin Lu · Tong He · Xuening Zhu · David Wipf

Title: Sparse Autoencoders, Again?

Abstract:
Is there really much more to say about sparse autoencoders (SAEs)? Autoencoders in general, and SAEs in particular, represent deep architectures that are capable of modeling lowdimensional latent structure in data. Such structure could reflect, among other things, correlation patterns in large language model activations, or complex natural image manifolds. And yet despite the wide-ranging applicability, there have been relatively few changes to SAEs beyond the original recipe from decades ago, namely, standard deep encoder/decoder layers trained with a classical/deterministic sparse regularizer applied within the latent space. One possible exception is the variational autoencoder (VAE), which adopts a stochastic encoder module capable of producing sparse representations when applied to manifold data. In this work we formalize underappreciated weaknesses with both canonical SAEs, as well as analogous VAEs applied to similar tasks, and propose a hybrid alternative model that circumvents these prior limitations. In terms of theoretical support, we prove that global minima of our proposed model recover certain forms of structured data spread across a union of manifolds. Meanwhile, empirical evaluations on synthetic and real-world datasets substantiate the efficacy of our approach in accurately estimating underlying manifold dimensions and producing sparser latent representations without compromising reconstruction error. In all cases, we are able to exceed the performance of equivalent-capacity SAEs and VAEs, as well as recent diffusion models where applicable.

Paperid:256

Authors:Andrea Cini · Alexander Jenkins · Danilo Mandic · Cesare Alippi · Filippo Maria Bianchi

Title: Relational Conformal Prediction for Correlated Time Series

Abstract:
We address the problem of uncertainty quantification in time series forecasting by exploiting observations at correlated sequences. Relational deep learning methods leveraging graph representations are among the most effective tools for obtaining point estimates from spatiotemporal data and correlated time series. However, the problem of exploiting relational structures to estimate the uncertainty of such predictions has been largely overlooked in the same context. To this end, we propose a novel distributionfree approach based on the conformal prediction framework and quantile regression. Despite the recent applications of conformal prediction to sequential data, existing methods operate independently on each target time series and do not account for relationships among them when constructing the prediction interval. We fill this void by introducing a novel conformal prediction method based on graph deep learning operators. Our method, named Conformal Relational Prediction (CoRel), does not require the relational structure (graph) to be known as a prior and can be applied on top of any pre-trained time series predictor. Additionally, CoRel includes an adaptive component to handle non-exchangeable data and changes in the input time series. Our approach provides accurate coverage and archives state-of-the-art uncertainty quantification in relevant benchmarks.

Paperid:257

Authors:Song Bian · Minghao Yan · Shivaram Venkataraman

Title: Scaling Inference-Efficient Language Models

Abstract:
Abstract:Scaling laws are powerful tools to predict the performance of large language models. However, current scaling laws fall short of accounting for inference costs. In this work, we first show that model architecture affects inference latency, where models of the same size can have up to $3.5\times$ difference in latency. To tackle this challenge, we modify the Chinchilla scaling laws to cooptimize the model parameter count, the number of training tokens, and the model architecture. Due to the reason that models of similar training loss exhibit gaps in downstream evaluation, we also propose a novel method to train inference-efficient models based on the revised scaling laws. We perform extensive empirical studies to fit and evaluate our inference-aware scaling laws. We vary model parameters from 80M to 1B, training tokens from 1.6B to 30B, and model shapes, training a total of 63 models. Guided by our inference-efficient scaling law and model selection method, we release the Morph-1B model, which improves inference latency by $1.8\times$ while maintaining accuracy on downstream tasks compared to open-source models, pushing the Pareto frontier of accuracy-latency tradeoff.

Paperid:258

Authors:Philippe Chlenski · Quentin Chu · Raiyan Khan · Kaizhu Du · Antonio Moretti · Itsik Pe'er

Title: Mixed-curvature decision trees and random forests

Abstract:
Decision trees (DTs) and their random forest (RF) extensions are workhorses of classification and regression in Euclidean spaces. However, algorithms for learning in nonEuclidean spaces are still limited. We extend DT and RF algorithms to product manifolds: Cartesian products of several hyperbolic, hyperspherical, or Euclidean components. Such manifolds handle heterogeneous curvature while still factorizing neatly into simpler components, making them compelling embedding spaces for complex datasets. Our novel angular reformulation of DTs respects the geometry of the product manifold, yielding splits that are geodesically convex, maximum-margin, and composable. In the special cases of single-component manifolds, our method simplifies to its Euclidean or hyperbolic counterparts, or introduces hyperspherical DT algorithms, depending on the curvature. We benchmark our method on various classification, regression, and link prediction tasks on synthetic data, graph embeddings, mixed-curvature variational autoencoder latent spaces, and empirical data. Compared to 7 other classifiers, product RFs ranked first on 25 out of 57 benchmarks, and placed in the top 2 for 46 out of 57. This highlights the value of product RFs as straightforward yet powerful new tools for data analysis in product manifolds.

Paperid:259

Authors:Yongqiang Chen · QUANMING YAO · Juzheng Zhang · James Cheng · Yatao Bian

Title: Hierarchical Graph Tokenization for Molecule-Language Alignment

Abstract:
Recent years have witnessed growing interest in extending the capabilities of large language models (LLMs) to molecular science. As LLMs are predominantly trained with text data, most existing approaches adopt a graph neural network to represent a molecule as a series of node tokens for moleculelanguage alignment, which, however, have overlooked the inherent hierarchical structures in molecules. Notably, high-order molecular structures contain rich semantics of functional groups, which encode crucial bio-chemical functionalities of the molecules. We show that neglecting the hierarchical information in tokenization will lead to subpar molecule-language alignment and severe hallucination. To address this limitation, we propose a novel strategy called HIerarchical GrapH Tokenization (HIGHT). HIGHT employs a hierarchical graph tokenizer that encodes the hierarchy of atom, motif, and molecular levels of informative tokens to improve the molecular perception of LLMs. HIGHT also adopts an augmented instruction tuning dataset, enriched with the hierarchical graph information, to further enhance the molecule-language alignment. Extensive experiments on 14 real-world benchmarks verify the effectiveness of HIGHT in reducing hallucination by 40%, and significant improvements in various molecule-language downstream tasks.

Paperid:260

Authors:Anvith Thudi · Evianne Rovers · Yangjun Ruan · Tristan Thrush · Chris Maddison

Title: MixMin: Finding Data Mixtures via Convex Minimization

Abstract:
Abstract:Modern machine learning pipelines are increasingly combining and mixing data from diverse and disparate sources, e.g., pretraining large language models. Yet, finding the optimal data mixture is a challenging and open problem. We formalize this data mixing problem as a bi-level objective: the best mixture is the one that would lead to the best model for a downstream objective. Unfortunately, this objective is generally intractable. In this paper, we make the observation that the bi-level data mixing objective becomes convex as our model class becomes larger. We develop and study a gradient-based approach for optimizing this convex objective, which we call MixMin, and test it on language modeling and chemistry tasks. MixMin was the only method that uniformly improved the data mixture in all our experiments. With MixMin, we improved the data mixture using less than 0.2% additional compute for a pythia-$410M$ model trained on $8.2B$ tokens, resulting between 1-5% relative improvement to negative log likelihood on PIQA, ARC Easy, SciQ, and OpenWebMath. Crucially, we found that MixMin mixtures for smaller models improved training of larger models, suggesting that MixMin mixtures may be scale-invariant. When mixing bioassay data to train an XGBoost model, we saw improvements to average precision scores of $0.03-0.15$.

Paperid:261

Authors:Jixian Zhou · Fang DONG · Ruijun Huang · Hengjie Cao · Mengyi Chen · Yifeng Yang · Anrui Chen · Mingzhi Dong · Yujiang Wang · Dongsheng Li · David Clifton · Qin Lv · Rui Zhu · Chun Zhang · Fan Yang · Tun Lu · Ning Gu · Li Shang

Title: Oracle-MoE: Locality-preserving Routing in the Oracle Space for Memory-constrained Large Lanuge Model Inference

Abstract:
Mixtureof-Experts (MoE) is widely adopted to deploy Large Language Models (LLMs) on edge devices with limited memory budgets.Although MoE is, in theory, an inborn memory-friendly architecture requiring only a few activated experts to reside in the memory for inference, current MoE architectures cannot effectively fulfill this advantage and will yield intolerable inference latencies of LLMs on memory-constrained devices. Our investigation pinpoints the essential cause as the remarkable temporal inconsistencies of inter-token expert activations, which generate overly frequent expert swapping demands dominating the latencies. To this end, we propose a novel MoE architecture, Oracle-MoE, to fulfill the real on-device potential of MoE-based LLMs. Oracle-MoE route tokens in a highly compact space suggested by attention scores, termed theoracle space, to effectively maintain the semantic locality across consecutive tokens to reduce expert activation variations, eliminating massive swapping demands. Theoretical analysis proves that Oracle-MoE is bound to provide routing decisions with better semantic locality and, therefore, better expert activation consistencies. Experiments on the pretrained GPT-2 architectures of different sizes (200M, 350M, 790M, and 2B) and downstream tasks demonstrate that without compromising task performance, our Oracle-MoE has achieved state-of-the-art inference speeds across varying memory budgets, revealing its substantial potential for LLM deployments in industry.

Paperid:262

Authors:Lele Fu · Bowen Deng · Sheng Huang · Tianchi Liao · Shirui Pan · Chuan Chen

Title: Less is More: Federated Graph Learning with Alleviating Topology Heterogeneity from A Causal Perspective

Abstract:
Federated graph learning (FGL) aims to collaboratively train a global graph neural network (GNN) on multiple private graphs with preserving the local data privacy. Besides the common cases of data heterogeneity in conventional federated learning, FGL faces the unique challenge of topology heterogeneity. Most of existing FGL methods alleviate the negative impact of heterogeneity by introducing global signals. However, the manners of creating increments might not be effective and significantly increase the computation amount. In light of this, we propose the FedATH, a FGL method with Alleviating Topology Heterogeneity from a causal perspective. Inspired by the causal theory, we argue that not all edges in a topology are necessary for the training objective, less topology information might make more sense. With the aid of edge evaluator, the local graphs are divided into causal and biased subgraphs. A dualGNN architecture is used to encode the two subgraphs into corresponding representations. Thus, the causal representations are drawn closer to the training objective while the biased representations are pulled away from it. Further, the Hilbert-Schmidt Independence Criterion is employed to strengthen the separability of the two subgraphs. Extensive experiments on six real-world graph datasets are conducted to demonstrate the superiority of the proposed FedATH over the compared approaches.

Paperid:263

Authors:Eric Zhao · Pranjal Awasthi · Sreenivas Gollapudi

Title: Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verification

Abstract:
Samplingbased search, a simple paradigm for utilizing test-time compute, involves generating multiple candidate responses and selecting the best one---typically by verifying each response for correctness. In this paper, we study the scaling trends governing sampling-based search. Among our findings is that simply scaling up a minimalist implementation that uses only random sampling and direct self-verification results in sustained performance improvements that, for example, elevate the Gemini v1.5 Pro model's reasoning capabilities past that of o1-Preview on popular benchmarks. We partially attribute the scalability of sampling-based search to a phenomenon of implicit scaling, where sampling a larger pool of responses in turn improves verification accuracy. We further identify two useful principles for improving self-verification capabilities with test-time compute: (1) comparing across responses provides helpful signals about the locations of errors and hallucinations, and (2) different model output styles are useful for different contexts---chains of thought are useful for reasoning but harder to verify. We also find that, though accurate verification can be elicited, frontier models demonstrate remarkably weak out-of-box verification capabilities and introduce a benchmark to measure progress on these deficiencies.

Paperid:264

Authors:Quentin Fruytier · Aryan Mokhtari · Sujay Sanghavi

Title: Learning Mixtures of Experts with EM: A Mirror Descent Perspective

Abstract:
Abstract:Classical Mixtures of Experts (MoE) are Machine Learning models that involve partitioning the input space, with a separate "expert" model trained on each partition. Recently, MoEbased model architectures have become popular as a means to reduce training and inference costs. There, the partitioning function and the experts are both learnt jointly via gradient descent-type methods on the log-likelihood. In this paper we study theoretical guarantees of the Expectation Maximization (EM) algorithm for the training of MoE models. We first rigorously analyze EM for MoE where the conditional distribution of the target and latent variable conditioned on the feature variable belongs to an exponential family of distributions and show its equivalence to projected Mirror Descent with unit step size and a Kullback-Leibler Divergence regularizer. This perspective allows us to derive new convergence results and identify conditions for local linear convergence; In the special case of mixture of $2$ linear or logistic experts, we additionally provide guarantees for linear convergence based on the signal-to-noise ratio. Experiments on synthetic and (small-scale) real-world data supports that EM outperforms the gradient descent algorithm both in terms of convergence rate and the achieved accuracy.

Paperid:265

Authors:Seung Lee · Hojoon Kim · Yutack Park · Dawoon Jeong · Seungwu Han · Yeonhong Park · Jae W. Lee

Title: FlashTP: Fused, Sparsity-Aware Tensor Product for Machine Learning Interatomic Potentials

Abstract:
Abstract:Machine Learning Interatomic Potentials (MLIPs) enable efficient molecular dynamics (MD) simulations with high accuracy. While equivariant MLIPs achieve stateof-the-art accuracy, they face significant computational bottlenecks centered around their Tensor-Product layer, which account for up to 75\% of training time and cause substantial memory overhead. We present FlashTP, a highly optimized tensor-product library that addresses these inefficiencies through kernel fusion, sparse computation, and path-aggregated execution. FlashTP achieves up to 29.1$\times$ and 24.3$\times$ kernel speedups over e3nn and NVIDIA cuEquivariance, respectively. For SevenNet-l3i5, it delivers 4.2$\times$ and 3.5$\times$ speedup while reducing peak memory usage by 6.3$\times$ and 6.2$\times$ for inference and training, respectively.

Paperid:266

Authors:Aya Kayal · Sattar Vakili · Laura Toni · Da-shan Shiu · Alberto Bernacchia

Title: Bayesian Optimization from Human Feedback: Near-Optimal Regret Bounds

Abstract:
Abstract:Bayesian optimization (BO) with preferencebased feedback has recently garnered significant attention due to its emerging applications. We refer to this problem as Bayesian Optimization from Human Feedback (BOHF), which differs from conventional BO by learning the best actions from a reduced feedback model, where only the preference between two actions is revealed to the learner at each time step. The objective is to identify the best action using a limited number of preference queries, typically obtained through costly human feedback. Existing work, which adopts the Bradley-Terry-Luce (BTL) feedback model, provides regret bounds for the performance of several algorithms. In this work, within the same framework we develop tighter performance guarantees. Specifically, we derive regret bounds of $\tilde{\mathcal{O}}(\sqrt{\Gamma(T)T})$, where $\Gamma(T)$ represents the maximum information gain—a kernel-specific complexity term—and $T$ is the number of queries. Our results significantly improve upon existing bounds. Notably, for common kernels, we show that the order-optimal sample complexities of conventional BO—achieved with richer feedback models—are recovered. In other words, the same number of preferential samples as scalar-valued samples is sufficient to find a nearly optimal solution.

Paperid:267

Authors:Jiajun Zhu · Peihao Wang · Ruisi Cai · Jason Lee · Pan Li · Zhangyang “Atlas” Wang

Title: Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding

Abstract:
Transformers rely on both contentbased and position-based addressing mechanisms to make predictions, but existing positional encoding techniques often diminish the effectiveness of position-based addressing. Many current methods enforce rigid patterns in attention maps, limiting the ability to model long-range dependencies and adapt to diverse tasks. Additionally, most positional encodings are learned as general biases, lacking the specialization required for different instances within a dataset. To address this, we propose conTextualized equivariAntPositionEncoding (TAPE), a novel framework that enhances positional embeddings by incorporating sequence content across layers. TAPE introduces dynamic, context-aware positional encodings, overcoming the constraints of traditional fixed patterns. By enforcing permutation and orthogonal equivariance, TAPE ensures the stability of positional encodings during updates, improving robustness and adaptability. Our method can be easily integrated into pre-trained transformers, offering parameter-efficient fine-tuning with minimal overhead. Extensive experiments show that TAPE achieves superior performance in language modeling, arithmetic reasoning, and long-context retrieval tasks compared to existing positional embedding techniques.

Paperid:268

Authors:Mohamed Ali Souibgui · Changkyu Choi · Andrey Barsky · Kangsoo Jung · Ernest Valveny · Dimosthenis Karatzas

Title: DocVXQA: Context-Aware Visual Explanations for Document Question Answering

Abstract:
We propose DocVXQA, a novel framework for visually selfexplainable document question answering, where the goal is not only to produce accurate answers to questions but also to learn visual heatmaps that highlight critical regions, offering interpretable justifications for the model decision. To integrate explanations into the learning process, we quantitatively formulate explainability principles as explicit learning criteria.Unlike conventional relevance map methods that solely emphasize regions relevant to the answer, our context-aware DocVXQA delivers explanations that are contextually sufficient yet representation-efficient. This fosters user trust while achieving a balance between predictive performance and interpretability in document visual question answering applications. Extensive experiments, including human evaluation, provide strong evidence supporting the effectiveness of our method.

Paperid:269

Authors:Chen Zhang · Weixin Bu · Zeyi Ren · Zhengwu Liu · Yik-Chung WU · Ngai Wong

Title: Nonparametric Teaching for Graph Property Learners

Abstract:
Inferring properties of graphstructured data,e.g., the solubility of molecules, essentially involves learning the implicit mapping from graphs to their properties. This learning process is often costly for graph property learners like Graph Convolutional Networks (GCNs). To address this, we propose a paradigm called Graph Nonparametric Teaching (GraNT) that reinterprets the learning process through a novel nonparametric teaching perspective. Specifically, the latter offers a theoretical framework for teaching implicitly defined (i.e., nonparametric) mappings via example selection. Such an implicit mapping is realized by a dense set of graph-property pairs, with the GraNT teacher selecting a subset of them to promote faster convergence in GCN training. By analytically examining the impact of graph structure on parameter-based gradient descent during training, and recasting the evolution of GCNs—shaped by parameter updates—through functional gradient descent in nonparametric teaching, we showfor the first timethat teaching graph property learners (i.e., GCNs) is consistent with teaching structure-aware nonparametric learners. These new findings readily commit GraNT to enhancing learning efficiency of the graph property learner, showing significant reductions in training time for graph-level regression (-36.62\%), graph-level classification (-38.47\%), node-level regression (-30.97\%) and node-level classification (-47.30\%), all while maintaining its generalization performance.

Paperid:270

Authors:Yuxuan Wang · Mingzhou Liu · Xinwei Sun · Wei Wang · Yizhou Wang

Title: Bayesian Active Learning for Bivariate Causal Discovery

Abstract:
Determining the direction of relationships between variables is fundamental for understanding complex systems across scientific domains. While observational data can uncover relationships between variables, it cannot distinguish between cause and effect without experimental interventions. To effectively uncover causality, previous works have proposed intervention strategies that sequentially optimize the intervention values. However, most of these approaches primarily maximized informationtheoretic gains that may not effectively measure the reliability of direction determination. In this paper, we formulate the causal direction identification as a hypothesis-testing problem, and propose a Bayes factor-based intervention strategy, which can quantify the evidence strength of one hypothesis (e.g., causal) over the other (e.g., non-causal). To balance the immediate and future gains of testing strength, we propose a sequential intervention objective over intervention values in multiple steps. By analyzing the objective function, we develop a dynamic programming algorithm that reduces the complexity from non-polynomial to polynomial. Experimental results on bivariate systems, tree-structured graphs, and an embodied AI environment demonstrate the effectiveness of our framework in direction determination and its extensibility to both multivariate settings and real-world applications.

Paperid:271

Authors:Jonggeon Park · Giung Nam · Hyunsu Kim · Jongmin Yoon · Juho Lee

Title: Ensemble Distribution Distillation via Flow Matching

Abstract:
Neural network ensembles have proven effective in improving performance across a range of tasks; however, their high computational cost limits their applicability in resourceconstrained environments or for large models. Ensemble distillation, the process of transferring knowledge from an ensemble teacher to a smaller student model, offers a promising solution to this challenge. The key is to ensure that the student model is both cost-efficient and achieves performance comparable to the ensemble teacher. With this in mind, we propose a novel ensemble distribution distillation method, which leverages flow matching to effectively transfer the diversity from the ensemble teacher to the student model. Our extensive experiments demonstrate the effectiveness of our proposed method compared to existing ensemble distillation approaches.

Paperid:272

Authors:Yu Sun · Xinhao Li · Karan Dalal · Jiarui Xu · Arjun Vikram · Genghan Zhang · Yann Dubois · Xinlei Chen · Xiaolong Wang · Sanmi Koyejo · Tatsunori Hashimoto · Carlos Guestrin

Title: Learning to (Learn at Test Time): RNNs with Expressive Hidden States

Abstract:
Selfattention performs well in long context but has quadratic complexity. Existing RNN layers have linear complexity, but their performance in long context is limited by the expressive power of their hidden state. Inspired by prior work (Schlag et al., 2021), we present a practical framework to instantiate sequence modeling layers with linear complexity and expressive hidden states. The key idea is to make the hidden state a machine learning model itself, and the update rule a step of self-supervised learning. Since the hidden state is updated by training even on test sequences, our layers are called Test-Time Training (TTT) layers. We consider two instantiations: TTT-Linear and TTT-MLP, whose hidden state is a linear model and a two-layer MLP respectively. We evaluate our instantiations at the scale of 125M to 1.3B parameters, comparing with a strong Transformer and Mamba, a modern RNN. Both TTT-Linear and TTT-MLP match or exceed the baselines. Similar to Transformer, they can keep reducing perplexity by conditioning on more tokens, while Mamba cannot after 16k context. With preliminary systems optimization, TTT-Linear is already faster than Transformer at 8k context and matches Mamba in wall-clock time. TTT-MLP still faces challenges in memory I/O, but shows larger potential in long context, pointing to a promising direction for future research.

Paperid:273

Authors:Sheheryar Mehmood · Peter Ochs

Title: Automatic Differentiation of Optimization Algorithms with Time-Varying Updates

Abstract:
Numerous optimization algorithms have a timevarying update rule thanks to, for instance, a changing step size, momentum parameter or, Hessian approximation. In this paper, we apply unrolled or automatic differentiation to a time-varying iterative process and provide convergence (rate) guarantees for the resulting derivative iterates. We adapt these convergence results and apply them to proximal gradient descent with variable step size and FISTA when solving partly-smooth problems. We test the convergence (rates) of these algorithms numerically through several experiments. Our theoretical and numerical results show that the convergence rate of the algorithm is reflected in its derivative iterates.

Paperid:274

Authors:Hyeong Kyu Choi · Maxim Khanov · Hongxin Wei · Sharon Li

Title: How Contaminated Is Your Benchmark? Quantifying Dataset Leakage in Large Language Models with Kernel Divergence

Abstract:
Dataset contamination, where evaluation datasets overlap with pretraining corpora, inflates performance metrics and undermines the reliability of model evaluations. Quantifying dataset contamination thus becomes essential to ensure that performance evaluations genuinely reflect a model's ability to generalize to unseen data, rather than relying on memorized examples. To address this problem, we propose Kernel Divergence Score (KDS), a novel method that quantifies dataset contamination by computing the divergence between the kernel similarity matrix of sample embeddings, before and after fine-tuning on the benchmark dataset. Leveraging the insight that fine-tuning affects unseen samples more significantly than seen ones, KDS provides a reliable measure of contamination. Through extensive experiments on controlled contamination scenarios, KDS demonstrates a near-perfect correlation with contamination levels and outperforms existing baselines. Additionally, we perform comprehensive ablation studies to analyze the impact of key design choices, providing deeper insights into the components and effectiveness of KDS. These ablations highlight the importance of leveraging fine-grained kernel-based information and confirm the reliability of the proposed framework across diverse datasets and settings.

Paperid:275

Authors:Shibo Jie · Yehui Tang · Kai Han · Yitong Li · Duyu Tang · Zhi-Hong Deng · Yunhe Wang

Title: Mixture of Lookup Experts

Abstract:
Mixtureof-Experts (MoE) activates only a subset of experts during inference, allowing the model to maintain low inference FLOPs and latency even as the parameter count scales up. However, since MoE dynamically selects the experts, all the experts need to be loaded into VRAM. Their large parameter size still limits deployment, and offloading, which load experts into VRAM only when needed, significantly increase inference latency. To address this, we propose Mixture of Lookup Experts (MoLoE), a new MoE architecture that is efficient in both communication and VRAM usage. In MoLoE, the experts are Feed-Forward Networks (FFNs) during training, taking the output of the embedding layer as input. Before inference, these experts can be re-parameterized as lookup tables (LUTs) that retrieves expert outputs based on input ids, and offloaded to storage devices. Therefore, we do not need to perform expert computations during inference. Instead, we directly retrieve the expert's computation results based on input ids and load them into VRAM, and thus the resulting communication overhead is negligible. Experiments show that, with the same FLOPs and VRAM usage, MoLoE achieves inference speeds comparable to dense models and significantly outperforms MoE with offloaded experts, while maintaining performance on par with MoE.

Paperid:276

Authors:Ren-Biao Liu · Anqi Li · ChaodingYang · Hui Sun · Ming Li

Title: Revisiting Chain-of-Thought in Code Generation: Do Language Models Need to Learn Reasoning before Coding?

Abstract:
Large Language Models (LLMs) have demonstrated exceptional performance in code generation, becoming increasingly vital for software engineering and development. Recently, Chainof-Thought (CoT) has proven effective for complex reasoning tasks by prompting LLMs to reason step-by-step and provide a final answer. However, research onhow LLMs learn to reason with CoT data for code generationremains limited. In this work, we revisit classic CoT training, which typically organizes reasoning steps before the final answer.We synthesize a dataset to distinguish the CoT process and code solution and then conduct extensive experiments to study how CoT works in code generation empirically.We observe counterfactual findings, indicating that the classic training paradigm does not yet benefit code generation. Instead, training LLMs to generate code first and then generate the CoT to explain reasoning steps during code generation is more effective.Specifically, our results indicate that a 9.9\% relative performance improvement can be achieved simply by changing the order between CoT and code.Our findings offer meaningful insights into leveraging CoT to improve the reasoning ability of code LLMs and enhance code generation.

Paperid:277

Authors:Muhammed Ustaomeroglu · Guannan Qu

Title: A Theoretical Study of (Hyper) Self-Attention through the Lens of Interactions: Representation, Training, Generalization

Abstract:
Abstract:Selfattention has emerged as a core component of modern neural architectures, yet its theoretical underpinnings remain elusive. In this paper, we study self-attention through the lens of *interacting entities*, ranging from agents in multi-agent reinforcement learning to alleles in genetic sequences, and show that a single layer linear self-attention can *efficiently* represent, learn, and generalize functions capturing pairwise interactions, including out-of-distribution scenarios. Our analysis reveals that self-attention acts as a *mutual interaction learner* under minimal assumptions on the diversity of interaction patterns observed during training, thereby encompassing a wide variety of real-world domains. In addition, we validate our theoretical insights through experiments demonstrating that self-attention, along with our proposed extensions, effectively learns interaction functions and generalizes across both population distributions and out-of-distribution scenarios. Extending self-attention, we introduce *HyperFeatureAttention*, a novel neural network module designed to learn *couplings of different feature-level interactions* between entities. Furthermore, we propose *HyperAttention*, a new mechanism that extends beyond pairwise interactions to capture multi-entity dependencies, such as three-way, four-way, or general $n$-way interactions.

Paperid:278

Authors:Ben Dai

Title: EnsLoss: Stochastic Calibrated Loss Ensembles for Preventing Overfitting in Classification

Abstract:
Empirical risk minimization (ERM) with a computationally feasible surrogate loss is a widely accepted approach for classification. Notably, the convexity and calibration (CC) properties of a loss function ensure consistency of ERM in maximizing accuracy, thereby offering a wide range of options for surrogate losses. In this article, we propose a novel ensemble method, namely EnsLoss, which extends the ensemble learning concept to combine loss functions within the ERM framework. A key feature of our method is the consideration on preserving the "legitimacy'' of the combined losses, i.e., ensuring the CC properties. Specifically, we first transform the CC conditions of losses into lossderivatives, thereby bypassing the need for explicit loss functions and directly generating calibrated loss-derivatives. Therefore, inspired by Dropout, EnsLoss enables loss ensembles through one training process with doubly stochastic gradient descent (i.e., random batch samples and random calibrated loss-derivatives). We theoretically establish the statistical consistency of our approach and provide insights into its benefits. The numerical effectiveness of EnsLoss compared to fixed loss methods is demonstrated through experiments on a broad range of 45 pairs of CIFAR10 datasets, the PCam image dataset, and 14 OpenML tabular datasets and with various deep learning architectures. Python repository and source code are available on anonymous GitHub at (https://anonymous.4open.science/r/ensLoss-53D1).

Paperid:279

Authors:Debangshu Banerjee · Tarun Suresh · Shubham Ugare · Sasa Misailovic · Gagandeep Singh

Title: CRANE: Expressive Grammar-Constrained LLM Generation

Abstract:
Code generation, symbolic math reasoning, and other tasks require LLMs to produce outputs that are both syntactically and semantically correct. Constrained LLM generation is a promising direction to enforce adherence to formal grammar, but prior works have empirically observed that strict enforcement of formal constraints often diminishes the reasoning capabilities of LLMs. In this work, we first provide a theoretical explanation for why constraining LLM outputs to very restrictive grammars that only allow syntactically valid final answers reduces the reasoning capabilities of the model. Second, we demonstrate that by augmenting the output grammar with carefully designed additional rules, it is always possible to preserve the reasoning capabilities of the LLM while ensuring syntactic and semantic correctness in its outputs. Building on these theoretical insights, we propose a reasoningaugmented constrained decoding algorithm, CRANE, which effectively balances the correctness of constrained generation with the flexibility of unconstrained generation. Experiments on multiple open-source LLMs and benchmarks show that CRANE significantly outperforms both state-of-the-art constrained decoding strategies and standard unconstrained decoding, showing up to a 9\% improvement over baselines on challenging symbolic mathematical benchmarks GSM-symbolic and FOLIO.

Paperid:280

Authors:EMANUELE SANSONE · Tim Lebailly · Tinne Tuytelaars

Title: Collapse-Proof Non-Contrastive Self-Supervised Learning

Abstract:
We present a principled and simplified design of the projector and loss function for noncontrastive self-supervised learning based on hyperdimensional computing. We theoretically demonstrate that this design introduces an inductive bias that encourages representations to be simultaneously decorrelated and clustered, without explicitly enforcing these properties. This bias provably enhances generalization and suffices to avoid known training failure modes, such as representation, dimensional, cluster, and intracluster collapses. We validate our theoretical findings on image datasets, including SVHN, CIFAR-10, CIFAR-100, and ImageNet-100. Our approach effectively combines the strengths of feature decorrelation and cluster-based self-supervised learning methods, overcoming training failure modes while achieving strong generalization in clustering and linear classification tasks.

Paperid:281

Authors:Abhijeet Mulgund · Chirag Pabbaraju

Title: Relating Misfit to Gain in Weak-to-Strong Generalization Beyond the Squared Loss

Abstract:
Abstract:The paradigm of weakto-strong generalization constitutes the training of a strong AI model on data labeled by a weak AI model, with the goal that the strong model nevertheless outperforms its weak supervisor on the target task of interest. For the setting of real-valued regression with the squared loss, recent work quantitatively characterizes the gain in performance of the strong model over the weak model in terms of the misfit between the strong and weak model. We generalize such a characterization to learning tasks whose loss functions correspond to arbitrary Bregman divergences when the strong class is convex. This extends the misfit-based characterization of performance gain in weak-to-strong generalization to classification tasks, as the cross-entropy loss can be expressed in terms of a Bregman divergence. In most practical scenarios, however, the strong model class may not be convex. We therefore weaken this assumption and study weak-to-strong generalization for convex combinations of $k$ strong models in the strong class, in the concrete setting of classification. This allows us to obtain a similar misfit-based characterization of performance gain, up to an additional error term that vanishes as $k$ gets large. Our theoretical findings are supported by thorough experiments on synthetic as well as real-world datasets.

Paperid:282

Authors:Guoqiang Zhang · John Lewis · W. Bastiaan Kleijn

Title: On Exact Bit-level Reversible Transformers Without Changing Architectures

Abstract:
Abstract:Various reversible deep neural networks (DNN) models have been proposed to reduce memory consumption in the training process. However, almost all existing reversible DNNs either require special nonstandard architectures or are constructed by modifying existing DNN architectures considerably to enable reversibility. In this work we present the BDIA-transformer, which is an exact bit-level reversible transformer that uses an unchanged standard architecture for inference. The basic idea is to first treat each transformer block as the Euler integration approximation for solving an ordinary differential equation (ODE) and then incorporate the technique of bidirectional integration approximation (BDIA) (originally designed for diffusion inversion) into the neural architecture, together with activation quantization to make it exactly bit-level reversible. In the training process, we let a hyper-parameter $\gamma$ in BDIA-transformer randomly take one of the two values $\{0.5, -0.5\}$ per training sample per transformer block for averaging every two consecutive integration approximations. As a result, BDIA-transformer can be viewed as training an ensemble of ODE solvers parameterized by a set of binary random variables, which regularizes the model and results in improved validation accuracy. Lightweight side information per transformer block is required to be stored in the forward process to account for binary quantization loss to enable exact bit-level reversibility. In the inference procedure, the expectation $\mathbb{E}(\gamma)=0$ is taken to make the resulting architectures of BDIA-transformer identical to transformers up to activation quantization. Our experiments in image classification, natural language generation, and language translation show that BDIA-transformers outperform their conventional counterparts significantly in terms of validation performance due to the regularization effect of the set of $\gamma$ random variables while also requiring considerably less training memory.

Paperid:283

Authors:Qiang Li · Michal Yemini · Hoi To Wai

Title: Clipped SGD Algorithms for Performative Prediction: Tight Bounds for Stochastic Bias and Remedies

Abstract:
This paper studies the convergence of clipped stochastic gradient descent (SGD) algorithms with decisiondependent data distribution. Our setting is motivated by privacy preserving optimization algorithms that interact with performative data where the prediction models can influence future outcomes. This challenging setting involves the non-smooth clipping operator and non-gradient dynamics due to distribution shifts. We make two contributions in pursuit for a performative stable solution with these algorithms. First, we characterize the stochastic bias with projected clipped SGD (PCSGD) algorithm which is caused by the clipping operator that prevents PCSGD from reaching a stable solution. When the loss function is strongly convex, we quantify the lower and upper bounds for this stochastic bias and demonstrate a bias amplification phenomenon with the sensitivity of data distribution. When the loss function is non-convex, we bound the magnitude of stationarity bias. Second, we propose remedies to mitigate the bias either by utilizing an optimal step size design for PCSGD, or to apply the recent DiceSGD algorithm [Zhang et al., 2024]. Our analysis is also extended to show that the latter algorithm is free from stochastic bias in the performative setting. Numerical experiments verify our findings.

Paperid:284

Authors:Wenbo Pan · Zhichao Liu · Qiguang Chen · Xiangyang Zhou · Yu Haining · Xiaohua Jia

Title: The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Analysis of Orthogonal Safety Directions

Abstract:
Large Language Models' safetyaligned behaviors, such as refusing harmful queries, can be represented by linear directions in activation space. Previous research modeled safety behavior with a single direction, limiting mechanistic understanding to an isolated safety feature. In this work, we discover that safety-aligned behavior is jointly controlled by multi-dimensional directions. Namely, we study the vector space of representation shifts during safety fine-tuning on Llama 3 8B for refusing jailbreaks. By studying orthogonal directions in the space, we first find that a dominant direction governs the model's refusal behavior, while multiple smaller directions represent distinct and interpretable features like hypothetical narrative and role-playing. We then measure how different directions promote or suppress the dominant direction, showing the important role of secondary directions in shaping the model's refusal representation. Finally, we demonstrate that removing certain trigger tokens in harmful queries can mitigate these directions to bypass the learned safety capability, providing new insights on understanding safety alignment vulnerability from a multi-dimensional perspective.

Paperid:285

Authors:GUOGUO AI · Guansong Pang · Hezhe Qiao · YuanGao · Hui Yan

Title: GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers

Abstract:
Graph Transformers (GTs) have demonstrated remarkable performance in graph representation learning over popular graph neural networks (GNNs). However, selfattention, the core module of GTs, preserves only low-frequency signals in graph features, leading to ineffectiveness in capturing other important signals like high-frequency ones. Some recent GT models help alleviate this issue, but their flexibility and expressiveness are still limited since the filters they learn are fixed on predefined graph spectrum or order. To tackle this challenge, we propose a Graph Fourier Kolmogorov-Arnold Transformer (GrokFormer), a novel GT model that learns highly expressive spectral filters with adaptive graph spectrum and order through a Fourier series modeling over learnable activation functions. We demonstrate theoretically and empirically that the proposed GrokFormer filter offers better expressiveness than other spectral methods. Comprehensive experiments on 10 real-world node classification datasets across various domains, scales, and graph properties, as well as 5 graph classification datasets, show that GrokFormer outperforms state-of-the-art GTs and GNNs. Our code is available at https://anonymous.4open.science/r/GrokFormer-1DCE.

Paperid:286

Authors:Feiyang Wang · Xingquan Zuo · Hai Huang · Gang Chen

Title: TtBA: Two-third Bridge Approach for Decision-Based Adversarial Attack

Abstract:
Abstract:A key challenge in blackbox adversarial attacks is the high query complexity in hard-label settings, where only the top-1 predicted label from the target deep model is accessible. In this paper, we propose a novel normal-vector-based method called Two-third Bridge Attack (TtBA). A innovative bridge direction is introduced which is a weighted combination of the current unit perturbation direction and its unit normal vector, controlled by a weight parameter $k$. We further use binary search to identify $k=k_\text{bridge}$, which has identical decision boundary as the current direction. Notably, we observe that $k=2/3 k_\text{bridge}$ yields a near-optimal perturbation direction, ensuring the stealthiness of the attack. In addition, we investigate the critical importance of local optima during the perturbation direction optimization process and propose a simple and effective approach to detect and escape such local optima. Experimental results on MNIST, FASHION-MNIST, CIFAR10, CIFAR100, and ImageNet datasets demonstrate the strong performance and scalability of our approach. Compared to state-of-the-art non-targeted and targeted attack methods, TtBA consistently delivers superior performance across most experimented datasets and deep learning models. Code is available at https://anonymous.4open.science/r/TtBA-6ECF.

Paperid:287

Authors:Yuefan Cao · Xiaoyu Li · Yingyiu Liang · Zhizhou Sha · Zhenmei Shi · Zhao Song · Jiahao Zhang

Title: Dissecting Submission Limit in Desk-Rejections: A Mathematical Analysis of Fairness in AI Conference Policies

Abstract:
As AI research surges in both impact and volume, conferences have imposed submission limits to maintain paper quality and alleviate organizational pressure. In this work, we examine the fairness of deskrejection systems under submission limits and reveal that existing practices can result in substantial inequities. Specifically, we formally define the paper submission limit problem and identify a critical dilemma: when the number of authors exceeds three, it becomes impossible to reject papers solely based on excessive submissions without negatively impacting innocent authors. Thus, this issue may unfairly affect early-career researchers, as their submissions may be penalized due to co-authors with significantly higher submission counts, while senior researchers with numerous papers face minimal consequences. To address this, we propose an optimization-based fairness-aware desk-rejection mechanism and formally define two fairness metrics: individual fairness and group fairness. We prove that optimizing individual fairness is NP-hard, whereas group fairness can be efficiently optimized via linear programming. Through case studies, we demonstrate that our proposed system ensures greater equity than existing methods, including those used in CVPR 2025, offering a more socially just approach to managing excessive submissions in AI conferences.

Paperid:288

Authors:Atefeh Gilani · Felipe Gomez · Shahab Asoodeh · Flavio Calmon · Oliver Kosut · Lalitha Sankar

Title: Optimizing Noise Distributions for Differential Privacy

Abstract:
Abstract:We propose a unified optimization framework for designing continuous and discrete noise distributions that ensure differential privacy (DP) by minimizing R\'enyi DP, a variant of DP, under a cost constraint. R\'enyi DP has the advantage that by considering different values of the R\'enyi parameter $\alpha$, we can tailor our optimization for any number of compositions. To solve the optimization problem, we reduce it to a finitedimensional convex formulation and perform preconditioned gradient descent. The resulting noise distributions are then compared to their Gaussian and Laplace counterparts. Numerical results demonstrate that our optimized distributions are consistently better, with significant improvements in $(\varepsilon, \delta)$-DP guarantees in the moderate composition regimes, compared to Gaussian and Laplace distributions with the same variance.

Paperid:289

Authors:Binchi Zhang · Zaiyi Zheng · Zhengzhang Chen · Jundong Li

Title: Beyond the Permutation Symmetry of Transformers: The Role of Rotation for Model Fusion

Abstract:
Symmetry in the parameter space of deep neural networks (DNNs) has proven beneficial for various deep learning applications. A wellknown example is the permutation symmetry in Multi-Layer Perceptrons (MLPs), where permuting the rows of weight matrices in one layer and applying the inverse permutation to adjacent layers yields a functionally equivalent model. While permutation symmetry fully characterizes the equivalence set for MLPs, its discrete nature limits its utility for transformers. In this paper, we introduce rotation symmetry, a novel form of parameter space symmetry for transformers that generalizes permutation symmetry by rotating parameter matrices in self-attention layers. Unlike permutation symmetry, rotation symmetry operates in a continuous domain, thereby significantly expanding the equivalence set for transformers. Based on this property, we propose a theoretically optimal parameter matching algorithm as a plug-and-play module to enhance model fusion. We evaluate our approach using pre-trained transformers across diverse natural language and vision tasks. Experimental results demonstrate that our rotation symmetry-based matching algorithm substantially improves model fusion, highlighting the potential of parameter space symmetry to facilitate model fusion. Our code is available on https://anonymous.4open.science/r/TransformerPermutation-8800.

Paperid:290

Authors:Alex Vitvitskyi · João Madeira Araujo · Marc Lackenby · Petar Veličković

Title: What makes a good feedforward computational graph?

Abstract:
As implied by the plethora of literature on graph rewiring, the choice of computational graph employed by a neural network can make a significant impact on its downstream performance. Certain effects related to the computational graph, such as underreaching and over-squashing, may even render the model incapable of learning certain functions. Most of these effects have only been thoroughly studied in the domain of undirected graphs; however, recent years have seen a significant rise in interest in feedforward computational graphs: directed graphs without any back edges. In this paper, we study the desirable properties of a feedforward computational graph, discovering two important complementary measures: fidelity and mixing time, and evaluating a few popular choices of graphs through the lens of these measures. Our study is backed by both theoretical analyses of the metrics' asymptotic behaviour for various graphs, as well as correlating these metrics to the performance of trained neural network models using the corresponding graphs.

Paperid:291

Authors:Ang Cao · Sergio Arnaud · Oleksandr Maksymets · Jianing Yang · Ayush Jain · Ada Martin · Vincent-Pierre Berges · Paul McVay · Ruslan Partsey · Aravind Rajeswaran · Franziska Meier · Justin Johnson · Jeong Joon Park · Alexander Sax

Title: LIFT-GS: Cross-Scene Render-Supervised Distillation for 3D Language Grounding

Abstract:
Our approach to training 3D visionlanguage understanding models is to train a feedforward model that makes predictions in 3D, but never requires 3D labels and is supervised only in 2D, using 2D losses and differentiable rendering.The approach is new for vision-language understanding. By treating the reconstruction as a ``latent variable'', we can render the outputs without placing unnecessary constraints on the network architecture (e.g. can be used with decoder-only models).For training, only need images and camera pose, and 2D labels. We show that we can even remove the need for 2D labels by using pseudo-labels from pretrained 2D models. We demonstrate this to pretrain a network, and we finetune it for 3D vision-language understanding tasks. We show this approach outperforms baselines/sota for 3D vision-language grounding, and also outperforms other 3D pretraining techniques.

Paperid:292

Authors:Ihab Bendidi · Yassir El Mesbahi · Alisandra Denton · Karush Suri · Kian Kenyon-Dean · Auguste Genovesio · Emmanuel Noutahi

Title: A Cross Modal Knowledge Distillation & Data Augmentation Recipe for Improving Transcriptomics Representations through Morphological Features

Abstract:
Understanding cellular responses to stimuli is crucial for biological discovery and drug development. Transcriptomics provides interpretable, genelevel insights, while microscopy imaging offers rich predictive features but is harder to interpret. Weakly paired datasets, where samples share biological states, enable multimodal learning but are scarce, limiting their utility for training and multimodal inference. We propose a framework to enhance transcriptomics by distilling knowledge from microscopy images. Using weakly paired data, our method aligns and binds modalities, enriching gene expression representations with morphological information. To address data scarcity, we introduce (1)Semi-Clipped, an adaptation of CLIP for cross-modal distillation using pretrained foundation models, achieving state-of-the-art results, and (2)PEA(PerturbationEmbeddingAugmentation), a novel augmentation technique that enhances transcriptomics data while preserving inherent biological information. These strategies improve the predictive power and retain the interpretability of transcriptomics, enabling rich unimodal representations for complex biological tasks.

Paperid:293

Authors:Thomas Massena · Léo Andéol · Thibaut Boissin · Franck Mamalet · Corentin FRIEDRICH · Mathieu Serrurier · Sébastien Gerchinovitz

Title: Efficient Robust Conformal Prediction via Lipschitz-Bounded Networks

Abstract:
Conformal Prediction (CP) has proven to be an effective posthoc method for improving the trustworthiness of neural networks by providing prediction sets with finite-sample guarantees. However, under adversarial attacks, classical conformal guarantees do not hold anymore: this problem is addressed in the field of Robust Conformal Prediction. Several methods have been proposed to provide robust CP sets with guarantees under adversarial perturbations, but, for large scale problems, these sets are either too large or the methods are too computationally demanding to be deployed in real life scenarios. In this work, we propose a new method that leverages Lipschitz-bounded networks to precisely and efficiently estimate robust CP sets. When combined with a 1-Lipschitz robust network, we demonstrate that our lip-rcp method outperforms state-of-the-art results in both the size of the robust CP sets and computational efficiency in medium and large-scale scenarios such as ImageNet. Taking a different angle, we also study vanilla CP under attack, and derive new worst-case coverage bounds of vanilla CP sets, which are valid simultaneously for all adversarial attack levels. Our lip-rcp method makes this second approach very efficient as well, as shown in the experiments.

Paperid:294

Authors:Zheng Gong · Ying Sun

Title: Outlier-Aware Post-training Quantization for Discrete Graph Diffusion Models

Abstract:
Abstract:Discrete Graph Diffusion Models (DGDMs) mark a pivotal advancement in graph generation, effectively preserving sparsity and structural integrity, thereby enhancing the learning of graph data distributions for diverse generative applications. Despite their potential, DGDMs are computationally intensive due to the numerous lowparameter yet high-computation operations, thereby increasing the need of inference acceleration. A promising solution to mitigate this issue is model quantization. However, existing quantization techniques for Image Diffusion Models (IDMs) face limitations in DGDMs due to differing diffusion processes, while Large Language Model (LLM) quantization focuses on reducing memory access latency of loading large parameters, unlike DGDMs, where inference bottlenecks are computations due to smaller model sizes. To fill this gap, we introduce Bit-DGDM, a post-training quantization framework for DGDMs which incorporates two novel ideas: (i) sparse-dense activation quantization sparsely modeling the activation outliers through adaptively selected, data-free thresholds in full-precision and quantizing the remaining to low-bit, and (ii) ill-conditioned low-rank decomposition decomposing the weights into low-rank component enable faster inference and an $\alpha$-sparsity matrix that models outliers. Extensive experiments demonstrate that Bit-DGDM not only reducing the memory usage from the FP32 baseline by up to $2.8\times$ and achieve up to $2.5\times$ speedup, but also achieve comparable performance to ultra-low precision of up to 4-bit.

Paperid:295

Authors:Yufei Guo · Yuhan Zhang · Zhou Jie · Xiaode Liu · Xin Tong · Yuanpei Chen · Weihang Peng · Zhe Ma

Title: ReverB-SNN: Reversing Bit of the Weight and Activation for Spiking Neural Networks

Abstract:
The Spiking Neural Network (SNN), a biologically inspired neural network infrastructure, has garnered significant attention recently. SNNs utilize binary spike activations for efficient information transmission, replacing multiplications with additions, thereby enhancing energy efficiency. However, binary spike activation maps often fail to capture sufficient data information, resulting in reduced accuracy.To address this challenge, we advocate reversing the bit of the weight and activation, called \textbf{ReverB}, inspired by recent findings that highlight greater accuracy degradation from quantizing activations compared to weights. Specifically, our method employs realvalued spike activations alongside binary weights in SNNs. This preserves the event-driven and multiplication-free advantages of standard SNNs while enhancing the information capacity of activations.Additionally, we introduce a trainable factor within binary weights to adaptively learn suitable weight amplitudes during training, thereby increasing network capacity. To maintain efficiency akin to vanilla \textbf{ReverB}, our trainable binary weight SNNs are converted back to standard form using a re-parameterization technique during inference.Extensive experiments across various network architectures and datasets, both static and dynamic, demonstrate that our approach consistently outperforms state-of-the-art methods.

Paperid:296

Authors:Harit Vishwakarma · Yi Chen · Satya Sai Srinath Namburi GNVV · Sui Jiet Tay · Ramya Vinayak · Frederic Sala

Title: Rethinking Confidence and Thresholds in Pseudolabeling-based SSL

Abstract:
Modern semisupervised learning (SSL) methods rely on pseudolabeling and consistency regularization. Pseudolabeling is typically performed by comparing the model's confidence scores and a predefined threshold. While several heuristics have been proposed to improve threshold selection, the underlying issues of overconfidence and miscalibration in confidence scores remain largely unaddressed, leading to inaccurate pseudolabels, degraded test accuracy, and prolonged training. We take a first-principles approach to learn confidence scores and thresholds with an explicit knob for error. This flexible framework addresses the fundamental question of optimal scores and threshold selection in pseudolabeling. Moreover, it gives practitioners a principled way to control the quality and quantity of pseudolabels. Such control is vital in SSL, where balancing pseudolabel quality and quantity directly affects model performance and training efficiency. Our experiments show that, by integrating this framework with modern SSL methods, we achieve significant improvements in accuracy and training efficiency. In addition, we provide novel insights on the trade-offs between the choices of the error parameter and the end model's performance.

Paperid:297

Authors:Ansgar Lößer · Max Schlecht · Florian Schintke · Joel Witzke · Matthias Weidlich · Björn Scheuermann

Title: Fast Min-$\epsilon$ Segmented Regression using Constant-Time Segment Merging

Abstract:
Abstract:Segmented regression is a statistical method that approximates a function $f$ by a piecewise function $\hat{f}$ using noisy data samples.*Min$\epsilon$* approaches aim to reduce the regression function's mean squared error (MSE) for a given number of $k$ segments.An optimal solution for *min-$\epsilon$* segmented regression is found in $\mathcal{O}(n^2)$ time (Bai & Perron, 1998; Yamamoto & Perron, 2013) for $n$ samples. For large datasets, current heuristics improve time complexity to $\mathcal{O}(n\log{n})$ (Acharya et al., 2016) but can result in large errors, especially when exactly $k$ segments are used.We present a method for *min-$\epsilon$* segmented regression that combines the scalability of top existing heuristic solutions with a statistical efficiency similar to the optimal solution. This is achieved by using a new method to merge an initial set of segments using precomputed matrices from samples, allowing both merging and error calculation in constant time.Our approach, using the same samples and parameter $k$, produces segments with up to 1,000 times lower MSE compared to Acharya et al. (2016) in about 100 times less runtime on data sets over $10^4$ samples.

Paperid:298

Authors:Yi Yu · Song Xia · SIYUAN YANG · Chenqi KONG · Wenhan Yang · Shijian Lu · Yap-peng Tan · Alex Kot

Title: MTL-UE: Learning to Learn Nothing for Multi-Task Learning

Abstract:
Most existing unlearnable strategies focus on preventing unauthorized users from training singletask learning (STL) models with personal data. Nevertheless, the paradigm has recently shifted towards multi-task data and multi-task learning (MTL), targeting generalist and foundation models that can handle multiple tasks simultaneously. Despite their growing importance, MTL data and models have been largely neglected while pursuing unlearnable strategies. This paper presents MTL-UE, the first unified framework for generating unlearnable examples for multi-task data and MTL models. Instead of optimizing perturbations for each sample, we design a generator-based structure that introduces label priors and class-wise feature embeddings which leads to much better attacking performance. In addition, MTL-UE incorporates intra-task and inter-task embedding regularization to increase inter-class separation and suppress intra-class variance which enhances the attack robustness greatly. Furthermore, MTL-UE is versatile with good supports for dense prediction tasks in MTL. It is also plug-and-play allowing integrating existing surrogate-dependent unlearnable methods with little adaptation. Extensive experiments show that MTL-UE achieves superior attacking performance consistently across 4 MTL datasets, 3 base UE methods, 5 model backbones, and 5 MTL task-weighting strategies.

Paperid:299

Authors:Linxi Zhao · Yihe Deng · Weitong Zhang · Quanquan Gu

Title: Mitigating Object Hallucination in Large Vision-Language Models via Image-Grounded Guidance

Abstract:
The advancement of Large VisionLanguage Models (LVLMs) has increasingly highlighted the critical issue of their tendency to hallucinate non-existing objects in the images. To address this issue, previous works focused on using specially curated datasets or powerful LLMs (e.g., GPT-3.5) to rectify the outputs of LVLMs. However, these approaches require either expensive training/fine-tuning or API access to advanced LLMs for post-generation correction. In response to these limitations, we propose Mitigating hallucinAtion via image-gRounded guIdaNcE (MARINE), a framework that is both training-free and API-free. MARINE effectively and efficiently reduces object hallucinations during inference by introducing image-grounded guidance to LVLMs. This is achieved by leveraging open-source vision models to extract object-level information, thereby enhancing the precision of LVLM-generated content. Our framework's flexibility further allows for the integration of multiple vision models, enabling more reliable and robust object-level guidance. Through comprehensive evaluations across popular LVLMs with diverse evaluation metrics and benchmarks, we demonstrate the effectiveness of MARINE, which even outperforms existing fine-tuning-based methods. Remarkably, it reduces hallucinations consistently in GPT-4V-assisted evaluation while maintaining the detailedness of LVLMs' generations.

Paperid:300

Authors:Laura Zheng · Wenjie Wei · Tony Wu · Jacob Clements · Shreelekha Revankar · Andre Harrison · Yu Shen · Ming Lin

Title: Adaptive Sensitivity Analysis for Robust Augmentation against Natural Corruptions in Image Segmentation

Abstract:
Achieving robustness in image segmentation models is challenging due to the finegrained nature of pixel-level classification. These models, which are crucial for many real-time perception applications, particularly struggle when faced with natural corruptions in the wild for autonomous systems. While sensitivity analysis can help us understand how input variables influence model outputs, its application to natural and uncontrollable corruptions in training data is computationally expensive. In this work, we present an adaptive, sensitivity-guided augmentation method to enhance robustness against natural corruptions. Our sensitivity analysis on average runs 10 times faster and requires about 200 times less storage than previous sensitivity analysis, enabling practical, on-the-fly estimation during training for a model-free augmentation policy. With minimal fine-tuning, our sensitivity-guided augmentation method achieves improved robustness on both real-world and synthetic datasets compared to state-of-the-art data augmentation techniques in image segmentation.

Paperid:301

Authors:Martino Bernasconi · Matteo Castiglioni · Andrea Celli

Title: No-Regret is not enough! Bandits with General Constraints through Adaptive Regret Minimization

Abstract:
Abstract:In the bandits with knapsacks framework (BwK) the learner has $m$ resourceconsumption (i.e., packing) constraints. We focus on the generalization of BwK in which the learner has a set of general long-term constraints. The goal of the learner is to maximize their cumulative reward, while at the same time achieving small cumulative constraints violations. In this scenario, there exist simple instances where conventional methods for BwK fail to yield sublinear violations of constraints. We show that it is possible to circumvent this issue by requiring the primal and dual algorithm to be weakly adaptive. Indeed, even without any information on the Slater's parameter $\rho$ characterizing the problem, the interaction between weakly adaptive primal and dual regret minimizers leads to a ``self-bounding'' behavior of dual variables. In particular, their norm remains suitably upper bounded across the entire time horizon even without explicit projection steps. By exploiting this property, we provide best-of-both-worlds guarantees for stochastic and adversarial inputs. In the first case, we show that the algorithm guarantees sublinear regret. In the latter case, we establish a tight competitive ratio of $\rho/(1+\rho)$. In both settings, constraints violations are guaranteed to be sublinear in time. Finally, this results allow us to obtain new result for the problem of contextual bandits with linear constraints, providing the first no-$\alpha$-regret guarantees for adversarial contexts.

Paperid:302

Authors:XINMIN QIU · Gege Chen · Bonan Li · Congying Han · Tiande Guo · Zicheng Zhang

Title: Feature out! Let Raw Image as Your Condition for Blind Face Restoration

Abstract:
Blind face restoration (BFR), which involves converting lowquality (LQ) images into high-quality (HQ) images, remains challenging due to complex and unknown degradations. While previous diffusion-based methods utilize feature extractors from LQ images as guidance, using raw LQ images directly as the starting point for the reverse diffusion process offers a theoretically optimal solution. In this work, we propose Pseudo-Hashing Image-to-image Schrödinger Bridge (P-I2SB), a novel framework inspired by optimal mass transport problems, which enhances the restoration potential of Schrödinger Bridge (SB) by correcting data distributions and effectively learning the optimal transport path between any two data distributions. Notably, we theoretically explore and identify that existing methods are limited by the optimality and reversibility of solutions in SB, leading to suboptimal performance. Our approach involves preprocessing HQ images during training by hashing them into pseudo-samples according to a rule related to LQ images, ensuring structural similarity in distribution. This guarantees optimal and reversible solutions in SB, enabling the inference process to learn effectively and allowing P-I2SB to achieve state-of-the-art results in BFR, with more natural textures and retained inference speed compared to previous methods.

Paperid:303

Authors:Daiheng Gao · Shilin Lu · Wenbo Zhou · Jiaming Chu · Jie Zhang · Mengxi Jia · Bang Zhang · Zhaoxin Fan · Weiming Zhang

Title: EraseAnything: Enabling Concept Erasure in Rectified Flow Transformers

Abstract:
Removing unwanted concepts from largescale text-to-image (T2I) diffusion models while maintaining their overall generative quality remains an open challenge. This difficulty is especially pronounced in emerging paradigms, such as Stable Diffusion (SD) v3 and Flux, which incorporate flow matching and transformer-based architectures. These advancements limit the transferability of existing concept-erasure techniques that were originally designed for the previous T2I paradigm (e.g., SD v1.4). In this work, we introduce EraseAnything, the first method specifically developed to address concept erasure within the latest flow-based T2I framework. We formulate concept erasure as a bi-level optimization problem, employing LoRA-based parameter tuning and an attention map regularizer to selectively suppress undesirable activations. Furthermore, we propose a self-contrastive learning strategy to ensure that removing unwanted concepts does not inadvertently harm performance on unrelated ones. Experimental results demonstrate that EraseAnything successfully fills the research gap left by earlier methods in this new T2I paradigm, achieving state-of-the-art performance across a wide range of concept erasure tasks.

Paperid:304

Authors:Maximilian Müller · Matthias Hein

Title: Mahalanobis++: Feature Normalization as the Missing Ingredient for Reliable OOD Detection

Abstract:
Abstract:Detecting outof-distribution (OOD) examples is an important task for deploying reliable machine learning models in safety-critial applications. While post-hoc methods based on the Mahalanobis distance applied to pre-logit features are among the most effective for ImageNet-scale OOD detection, their performance varies significantly across models. We connect this inconsistency to strong variations in feature norms, indicating severe violations of the Gaussian assumption underlying the Mahalanobis distance estimation. We show that simple $\ell_2$-normalization of the features mitigates this problem effectively, aligning better with the premise of normally distributed data with shared covariance matrix. Extensive experiments on 44 models across diverse architectures and pretraining schemes show that $\ell_2$-normalization improves the conventional Mahalanobis distance-based approaches significantly and consistently, and outperforms other recently proposed OOD detection methods.

Paperid:305

Authors:Guanghui Wang · Zhiyong Yang · Zitai Wang · Shi Wang · Qianqian Xu · Qingming Huang

Title: ABKD: Pursuing a Proper Allocation of the Probability Mass in Knowledge Distillation via $\alpha$-$\beta$-Divergence

Abstract:
Abstract:Knowledge Distillation (KD) transfers knowledge from a large teacher model to a smaller student model by minimizing the divergence between their output distributions, typically using forward KullbackLeibler divergence (FKLD) or reverse KLD (RKLD). It has become an effective training paradigm due to the broader supervision information provided by the teacher distribution compared to one-hot labels. We identify that the core challenge in KD lies in balancing two mode-concentration effects: the \textbf{\textit{Hardness-Concentration}} effect, which refers to focusing on modes with large errors, and the \textbf{\textit{Confidence-Concentration}} effect, which refers to focusing on modes with high student confidence. Through an analysis of how probabilities are reassigned during gradient updates, we observe that these two effects are entangled in FKLD and RKLD, but in extreme forms. Specifically, both are too weak in FKLD, causing the student to fail to concentrate on the target class. In contrast, both are too strong in RKLD, causing the student to overly emphasize the target class while ignoring the broader distributional information from the teacher. To address this imbalance, we propose ABKD, a generic framework with $\alpha$-$\beta$-divergence. Our theoretical results show that ABKD offers a smooth interpolation between FKLD and RKLD, achieving a better trade-off between these effects. Extensive experiments on 17 language/vision datasets with 12 teacher-student settings confirm its efficacy.

Paperid:306

Authors:Yiding Chen · Yiyi Zhang · Owen Oertell · Wen Sun

Title: Convergence of Consistency Model with Multistep Sampling under General Data Assumptions

Abstract:
Diffusion models accomplish remarkable success in data generation tasks across various domains. However, the iterative sampling process is computationally expensive. Consistency models are proposed to learn consistency functions to map from noise to data directly, which allows onestep fast data generation and multistep sampling to improve sample quality. In this paper, we study the convergence of consistency models when the self-consistency property holds approximately under the training distribution. Our analysis requires only mild data assumption and applies to a family of forward processes. When the target data distribution has bounded support or has tails that decay sufficiently fast, we show that the samples generated by the consistency model are close to the target distribution in Wasserstein distance; when the target distribution satisfies some smoothness assumption, we show that with an additional perturbation step for smoothing, the generated samples are close to the target distribution in total variation distance. We provide two case studies with commonly chosen forward processes to demonstrate the benefit of multistep sampling.

Paperid:307

Authors:Shiba Biswal · Karthik Elamvazhuthi · Rishi Sonthalia

Title: Universal Approximation of Mean-Field Models via Transformers

Abstract:
Abstract:This paper investigates the use of transformers to approximate the meanfield dynamics of interacting particle systems exhibiting collective behavior. Such systems are fundamental in modeling phenomena across physics, biology, and engineering, including opinion formation, biological networks, and swarm robotics. The key characteristic of these systems is that the particles are indistinguishable, leading to permutation-equivariant dynamics. First, we empirically demonstrate that transformers are well-suited for approximating a variety of mean field models, including the Cucker-Smale model for flocking and milling, and the mean-field system for training two-layer neural networks. We validate our numerical experiments via mathematical theory. Specifically, we prove that if a finite-dimensional transformer effectively approximates the finite-dimensional vector field governing the particle system, then the $L_\infty$ distance between the \textit{expected transformer} and the infinite-dimensional mean-field vector field can be bounded by a function of the number of particles observed during training. Leveraging this result, we establish theoretical bounds on the distance between the true mean-field dynamics and those obtained using the transformer.

Paperid:308

Authors:Bryan Lincoln Marques de Oliveira · Brandão · Murilo L. da Luz · Luana Martins · telma de lima soares · Luckeciano Melo

Title: Sliding Puzzles Gym: A Scalable Benchmark for State Representation in Visual Reinforcement Learning

Abstract:
Learning effective visual representations enables agents to extract meaningful information from raw sensory inputs, which is essential for generalizing across different tasks. However, evaluating representation learning separately from policy learning remains a challenge with most reinforcement learning (RL) benchmarks. To address this gap, we introduce the Sliding Puzzles Gym (SPGym), a novel benchmark that reimagines the classic 8tile puzzle with a visual observation space of images sourced from arbitrarily large datasets. SPGym provides precise control over representation complexity through visual diversity, allowing researchers to systematically scale the representation learning challenge while maintaining consistent environment dynamics. Despite the apparent simplicity of the task, our experiments with both model-free and model-based RL algorithms reveal fundamental limitations in current methods. As we increase visual diversity by expanding the pool of possible images, all tested algorithms show significant performance degradation, with even state-of-the-art methods struggling to generalize across different visual inputs while maintaining consistent puzzle-solving capabilities. These results highlight critical gaps in visual representation learning for RL and provide clear directions for improving robustness and generalization in decision-making systems.

Paperid:309

Authors:Pengchong Hu · Zhizhong Han

Title: VTGaussian-SLAM: RGBD SLAM for Large Scale Scenes with Splatting View-Tied 3D Gaussians

Abstract:
Jointly estimating camera poses and mapping scenes are fundamental tasks in simultaneous localization and mapping (SLAM) from RGBD images. Stateof-the-art methods employ 3D Gaussians to represent a scene, and render these Gaussians through splatting for higher efficiency and better rendering. However, these methods cannot scale up to extremely large scenes, due to the inefficient tracking and mapping strategies that need to optimize all 3D Gaussisans in the limited GPU memories throughout the training to maintain the geometry and color consistency to all previous RGBD observations. To resolve this issue, we propose novel tracking and mapping strategies to work with a novel 3D representation, dubbed view-tied 3D Gaussians, for RGBD SLAM systems. View-tied 3D Gaussians is a kind of simplified Gaussians, which is tied to depth pixels, without needing to learn locations, rotations, and multi-dimensional variances. Tying Gaussians to views not only significantly saves storage but also allows us to employ many more Gaussians to represent local details in the limited GPU memories. Moreover, our strategies remove the need of maintaining all Guassians learnable throughout the training, improving rendering quality and tracking and mapping accuracy. We justify the effectiveness of these designs, and report better performance over the latest methods on the widely used benchmarks in terms of rendering and tracking accuracy and scalability.

Paperid:310

Authors:Matthew Wicker · Philip Sosnin · Igor Shilov · Adrianna Janik · Mark Müller · Yves-Alexandre de Montjoye · Adrian Weller · Calvin Tsay

Title: Certification for Differentially Private Prediction in Gradient-Based Training

Abstract:
We study private prediction where differential privacy is achieved by adding noise to the outputs of a nonprivate model. Existing methods rely on noise proportional to the global sensitivity of the model, often resulting in sub-optimal privacy-utility trade-offs compared to private training. We introduce a novel approach for computing dataset-specific upper bounds on prediction sensitivity by leveraging convex relaxation and bound propagation techniques. By combining these bounds with the smooth sensitivity mechanism, we significantly improve the privacy analysis of private prediction compared to global sensitivity-based approaches. Experimental results across real-world datasets in medical image classification and natural language processing demonstrate that our sensitivity bounds are can be orders of magnitude tighter than global sensitivity. Our approach provides a strong basis for the development of novel privacy preserving technologies.

Paperid:311

Authors:David K Park · Xihaier Luo · Guang Zhao · Seungjun Lee · Miruna Oprescu · Shinjae Yoo

Title: SCENT: Robust Spatiotemporal Learning for Continuous Scientific Data via Scalable Conditioned Neural Fields

Abstract:
Spatiotemporal learning is challenging due to the intricate interplay between spatial and temporal dependencies, the high dimensionality of the data, and scalability constraints. These challenges are further amplified in scientific domains, where data is often irregularly distributed (e.g., missing values from sensor failures) and highvolume (e.g., high-fidelity simulations), posing additional computational and modeling difficulties. In this paper, we present SCENT, a novel framework for scalable and continuity-informed spatiotemporal representation learning. SCENT unifies interpolation, reconstruction, and forecasting within a single architecture. Built on a transformer-based encoder-processor-decoder backbone, SCENT introduces learnable queries to enhance generalization and a query-wise cross-attention mechanism to effectively capture multi-scale dependencies. To ensure scalability in both data size and model complexity, we incorporate a sparse attention mechanism, enabling flexible output representations and efficient evaluation at arbitrary resolutions. We validate SCENT through extensive simulations and real-world experiments, demonstrating state-of-the-art performance across multiple challenging tasks while achieving superior scalability.

Paperid:312

Authors:Piotr Kubaty · Bartosz Wójcik · Bartłomiej Krzepkowski · Monika Michaluk · Tomasz Trzcinski · Jary Pomponi · Kamil Adamczewski

Title: How to Train Your Multi-Exit Model? Analyzing the Impact of Training Strategies

Abstract:
Early exits enable the network's forward pass to terminate early by attaching trainable internal classifiers to the backbone network. Existing earlyexit methods typically adopt either a joint training approach, where the backbone and exit heads are trained simultaneously, or a disjoint approach, where the heads are trained separately. However, the implications of this choice are often overlooked, with studies typically adopting one approach without adequate justification. This choice influences training dynamics and its impact remains largely unexplored. In this paper, we introduce a set of metrics to analyze early-exit training dynamics and guide the choice of training strategy. We demonstrate that conventionally used joint and disjoint regimes yield suboptimal performance. To address these limitations, we propose a mixed training strategy: the backbone is trained first, followed by the training of the entire multi-exit network. Through comprehensive evaluations of training strategies across various architectures, datasets, and early-exit methods we present strengths and weaknesses of the early exit training strategies. In particular, we show consistent improvements in performance and efficiency using the proposed mixed strategy.

Paperid:313

Authors:Amber Xie · Oleh Rybkin · Dorsa Sadigh · Chelsea Finn

Title: Latent Diffusion Planning for Imitation Learning

Abstract:
Recent progress in imitation learning has been enabled by policy architectures that scale to complex visuomotor tasks, multimodal distributions, and large datasets. However, these methods often rely on learning from large amount of expert demonstrations.To address these shortcomings, we propose Latent Diffusion Planning (LDP), a modular approach consisting of a planner which can leverage actionfree demonstrations, and an inverse dynamics model which can leverage suboptimal data, that both operate over a learned latent space. First, we learn a compact latent space through a variational autoencoder, enabling effective forecasting of future states in image-based domains. Then, we train a planner and an inverse dynamics model with diffusion objectives. By separating planning from action prediction, LDP can benefit from the denser supervision signals of suboptimal and action-free data.On simulated visual robotic manipulation tasks, LDP outperforms state-of-the-art imitation learning approaches, as they cannot leverage such additional data.

Paperid:314

Authors:Shuqing Luo · Pingzhi Li · Jie Peng · Yang Zhao · Yu Cao · Yu Cheng · Tianlong Chen

Title: Occult: Optimizing Collaborative Communications across Experts for Accelerated Parallel MoE Training and Inference

Abstract:
Abstract:Mixtureof-experts (MoE) architectures could achieve impressive computational efficiency with expert parallelism, which relies heavily on all-to-all communication across devices. Unfortunately, such communication overhead typically constitutes a significant portion of the total runtime, hampering the scalability of distributed training and inference for modern MoE models (consuming over 40% runtime in large-scale training). In this paper, we first define $\textit{collaborative communication}$ to illustrate this intrinsic limitation, and then propose system- and algorithm-level innovations to reduce communication costs. Specifically, given a pair of experts co-activated by one token, we call them as $\textit{collaborated}$, which comprises $2$ cases as $\textit{intra-}$ and $\textit{inter-collaboration}$, depending on whether they are kept on the same device. Our pilot investigations reveal that augmenting the proportion of intra-collaboration can accelerate expert parallel at scale. It motivates us to strategically $\underline{\texttt{o}}$ptimize $\underline{\texttt{c}}$ollaborative $\underline{\texttt{c}}$omm$\underline{\texttt{u}}$nication for acce$\underline{\texttt{l}}$era$\underline{\texttt{t}}$ed MoE training and inference, dubbed $\textbf{\texttt{Occult}}$. Our designs are capable of $\underline{either}$ delivering exact results with reduced communication cost, $\underline{or}$ controllably minimizing the cost with collaboration pruning, materialized by modified fine-tuning. Comprehensive experiments on various MoE-LLMs demonstrate that $\texttt{Occult}$ can be faster than popular state-of-the-art inference or training frameworks (over 50% speed up across multiple tasks and models) with comparable or superior quality compared to the standard fine-tuning. Codes will be available upon acceptance.

Paperid:315

Authors:Flavio Petruzzellis · Cristina Cornelio · Pietro Lió

Title: Hierarchical Planning for Complex Tasks with Knowledge Graph-RAG and Symbolic Verification

Abstract:
Large Language Models (LLMs) have shown promise as robotic planners but often struggle with longhorizon and complex tasks, especially in specialized environments requiring external knowledge. While hierarchical planning and Retrieval-Augmented Generation (RAG) address some of these challenges, they remain insufficient on their own and a deeper integration is required for achieving more reliable systems. To this end, we propose a neuro-symbolic approach that enhances LLMs-based planners with Knowledge Graph-based RAG for hierarchical plan generation. This method decomposes complex tasks into manageable subtasks, further expanded into executable atomic action sequences. To ensure formal correctness and proper decomposition, we integrate a symbolic validator, which also functions as a failure detector by aligning expected and observed world states. Our evaluation against baseline methods demonstrates the consistent significant advantages of integrating hierarchical planning, symbolic verification, and RAG across tasks of varying complexity and different LLMs. Additionally, our experimental setup and novel metrics not only validates our approach for complex planning but also serves as a tool for assessing LLMs' reasoning and compositional capabilities.

Paperid:316

Authors:Sunil Madhow · Dheeraj Baby · Yu-Xiang Wang

Title: AKORN: Adaptive Knots generated Online for RegressioN Splines

Abstract:
Abstract:In order to attain optimal rates, stateof-the-art algorithms for non-parametric regression require that a hyperparameter be tuned according to the smoothness of the ground truth (Tibshirani, 2014). This amounts to an assumption of oracle access to certain features of the data-generating process. We present a parameter-free algorithm for offline non-parametric regression over $TV_1$-bounded functions. By feeding offline data into an optimal online denoising algorithm styled after (Baby et al., 2021), we are able to use change-points to adaptively select knots that respect the geometry of the underlying ground truth. We call this procedure AKORN (Adaptive Knots gener- ated Online for RegressioN splines). By combining forward and backward passes over the data, we obtain an estimator whose empirical performance is close to Trend Filtering (Kim et al., 2009; Tibshirani, 2014), even when we provide the latter with oracle knowledge of the ground truth’s smoothness.

Paperid:317

Authors:Lokesh Veeramacheneni · Moritz Wolter · Hilde Kuehne · Juergen Gall

Title: Canonical Rank Adaptation: An Efficient Fine-Tuning Strategy for Vision Transformers

Abstract:
Modern methods for finetuning a Vision Transformer (ViT) like Low-Rank Adaptation (LoRA) and its variants demonstrate impressive performance. However, these methods ignore the high-dimensional nature of Multi-Head Attention (MHA) weight tensors. To address this limitation, we propose Canonical Rank Adaptation (CaRA). CaRA leverages tensor mathematics, first by tensorising the transformer into two different tensors; one for projection layers in MHA and the other for feed-forward layers. Second, the tensorised formulation is fine-tuned using the low-rank adaptation in Canonical-Polyadic Decomposition (CPD) form. Employing CaRA efficiently minimizes the number of trainable parameters. Experimentally, CaRA outperforms existing Parameter-Efficient Fine-Tuning (PEFT) methods in visual classification benchmarks such as Visual Task Adaptation Benchmark (VTAB)-1k and Fine-Grained Visual Categorization (FGVC).

Paperid:318

Authors:Suryanarayana Sankagiri · Jalal Etesami · Matthias Grossglauser

Title: Recommendations with Sparse Comparison Data: Provably Fast Convergence for Nonconvex Matrix Factorization

Abstract:
This paper provides a theoretical analysis of a new learning problem for recommender systems where users provide feedback by comparing pairs of items instead of rating them individually. We assume that comparisons stem from latent user and item features, which reduces the task of predicting preferences to learning these features from comparison data. Similar to the classical matrix completion problem, the main challenge in this learning task is that the resulting loss function is nonconvex. Our analysis shows that the loss function exhibits (restricted) strong convexity near the true solution, which ensures gradientbased methods converge exponentially, given a warm start. Importantly, this result holds in a sparse data regime, where each user compares only a few pairs of items. Our main technical contribution is to extend certain concentration inequalities commonly used in matrix completion to our model. Our work demonstrates that learning personalized recommendations from comparison data is computationally and statistically efficient.

Paperid:319

Authors:Ariel Procaccia · Benjamin Schiffer · Shirley Zhang

Title: Clone-Robust AI Alignment

Abstract:
A key challenge in training Large Language Models (LLMs) is properly aligning them with human preferences. Reinforcement Learning with Human Feedback (RLHF) uses pairwise comparisons from human annotators to train reward functions and has emerged as a popular alignment method. However, input datasets in RLHF can be unbalanced due to adversarial manipulation or inadvertent repetition. Therefore, we want RLHF algorithms to perform well even when the set of alternatives is not uniformly distributed. Drawing on insights from social choice theory, we introduce robustness to approximate clones, a desirable property of RLHF algorithms which requires that adding nearduplicate alternatives does not significantly change the learned reward function. We first demonstrate that the standard RLHF algorithm based on regularized maximum likelihood estimation (MLE) fails to satisfy this property. We then propose the weighted MLE, a new RLHF algorithm that modifies the standard regularized MLE by weighting alternatives based on their similarity to other alternatives. This new algorithm guarantees robustness to approximate clones while preserving desirable theoretical properties.

Paperid:320

Authors:Laure Ciernik · Lorenz Linhardt · Marco Morik · Jonas Dippel · Simon Kornblith · Lukas Muttenthaler

Title: Objective drives the consistency of representational similarity across datasets

Abstract:
The Platonic Representation Hypothesis claims that recent foundation models are converging to a shared representation space as a function of their downstream task performance, irrespective of the objectives and data modalities used to train these models. Representational similarity is generally measured for individual datasets and is not necessarily consistent across datasets. Thus, one may wonder whether this convergence of model representations is confounded by the datasets commonly used in machine learning. Here, we propose a systematic way to measure how representational similarity between models varies with the set of stimuli used to construct the representations. We find that the objective function is a crucial factor in determining the consistency of representational similarities across datasets. Specifically, selfsupervised vision models learn representations whose relative pairwise similarities generalize better from one dataset to another compared to those of image classification or image-text models. Moreover, the correspondence between representational similarities and the models' task behavior is dataset-dependent, being most strongly pronounced for single-domain datasets. Our work provides a framework for analyzing similarities of model representations across datasets and linking those similarities to differences in task behavior.

Paperid:321

Authors:Erica Zhang · Fangzhao Zhang · Mert Pilanci

Title: Active Learning of Deep Neural Networks via Gradient-Free Cutting Planes

Abstract:
Active learning methods aim to improve sample complexity in machine learning. In this work, we investigate an active learning scheme via a novel gradientfree cutting-plane training method for ReLU networks of arbitrary depth and develop a convergence theory. We demonstrate, for the first time, that cutting-plane algorithms, traditionally used in linear models, can be extended to deep neural networks despite their nonconvexity and nonlinear decision boundaries. Moreover, this training method induces the first deep active learning scheme known to achieve convergence guarantees, revealing a geometric contraction rate of the feasible set. We exemplify the effectiveness of our proposed active learning method against popular deep active learning baselines via both synthetic data experiments and sentimental classification task on real datasets.

Paperid:322

Authors:Sidak Pal Singh · Hossein Mobahi · Atish Agarwala · Yann Nicolas Dauphin

Title: Avoiding spurious sharpness minimization broadens applicability of SAM

Abstract:
Curvature regularization techniques like Sharpness Aware Minimization (SAM) have shown great promise in improving generalization on vision tasks. However, we find that SAM performs poorly in domains like natural language processing (NLP), often degrading performance even with twice the compute budget. We investigate the discrepancy across domains and find that in the NLP setting, SAM is dominated by regularization of the logit statistics --- instead of improving the geometry of the function itself. We use this observation to develop an alternative algorithm we call Functional SAM, which regularizes curvature only through modification of the statistics of the overall function implemented by the neural network, and avoids spurious minimization through logit manipulation. Furthermore, we argue that preconditioning the SAM perturbation also prevents spurious minimization, and when combined with Functional SAM, it gives further improvements. Our proposed algorithms show improved performance over AdamW and SAM baselines when trained for an equal number of steps, in both fixed-length and Chinchilla-style training settings, at various model scales (including billion-parameter scale). On the whole, our work highlights the importance of more precise characterizations of sharpness in broadening the applicability of curvature regularization to large language models (LLMs)

Paperid:323

Authors:Yunfeng Liao · Jiawen Guan · Xiucheng Li

Title: Curvature-aware Graph Attention for PDEs on Manifolds

Abstract:
Deep models have recently achieved remarkable performances in solving partial differential equations (PDEs). The previous methods are mostly focused on PDEs arising in Euclidean spaces with less emphasis on the general manifolds with rich geometry. Several proposals attempt to account for the geometry by exploiting the spatial coordinates but overlook the underlying intrinsic geometry of manifolds. In this paper, we propose a Curvatureaware Graph Attention for PDEs on manifolds by exploring the important intrinsic geometric quantities such as curvature and discrete gradient operator. It is realized via parallel transport and tensor field on manifolds. To accelerate computation, we present three curvature-oriented graph embedding approaches and derive closed-form parallel transport equations, and a sub-tree partition method is also developed to promote parameter-sharing. Our proposed curvature-aware attention can be used as a replacement for vanilla attention, and experiments show that it significantly improves the performance of the existing methods for solving PDEs on manifolds. Our code is available at https://anonymous.4open.science/r/icml2025-5376/.

Paperid:324

Authors:Ayça Takmaz · Cristiano Saltori · Neehar Peri · Tim Meinhardt · Riccardo de Lutio · Laura Leal-Taixé · Aljosa Osep

Title: Towards Learning to Complete Anything in Lidar

Abstract:
Panoptic scene completion methods estimate dense object and scene geometry from sparse Lidar data. However, contemporary methods can only complete and recognize objects from a closed vocabulary, labeled in existing Lidar datasets. To overcome this limitation, we introduce CAL (Complete Anything in Lidar). Our zeroshot approach uses the temporal context from multi-modal sensor sequences to mine object shapes and semantic features that are then distilled into a Lidar-only instance-level completion and recognition model. Although we only mine partial shape completions, we find that our distilled model learns to infer full object shapes from multiple such partial observations across the dataset. We show that our model can be prompted on standard benchmarks for Semantic and Panoptic Scene Completion, localize objects as (amodal) 3D bounding boxes, and recognize objects beyond fixed class vocabularies.

Paperid:325

Authors:Louis Béthune · David Grangier · Dan Busbridge · Eleonora Gualdoni · Marco Cuturi · Pierre Ablin

Title: Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection

Abstract:
A widespread strategy to obtain a language model that performs well on a target domain is to finetune a pretrained model to perform unsupervised nexttoken prediction on data from that target domain.Finetuning presents two challenges:(i)if the amount of target data is limited, as in most practical applications, the model will quickly overfit, and(ii)the model will drift away from the original model, forgetting the pretraining data and the generic knowledge that comes with it.Our goal is to derive scaling laws that quantify these two phenomena for various target domains, amounts of available target data, and model scales.We measure the efficiency of injecting pretraining data into the finetuning data mixture to avoid forgetting and mitigate overfitting.A key practical takeaway from our study is that injecting as little as 1% of pretraining data in the finetuning data mixture prevents the model from forgetting the pretraining set.

Paperid:326

Authors:Ekaterina Borodich · Alexander Gasnikov · Dmitry Kovalev

Title: On Linear Convergence in Smooth Convex-Concave Bilinearly-Coupled Saddle-Point Optimization: Lower Bounds and Optimal Algorithms

Abstract:
Abstract:We revisit the smooth convexconcave bilinearly-coupled saddle-point problem of the form $\min_x\max_y f(x) + \langle y,\mathbf{B} x\rangle - g(y)$. In the highly specific case where function $f(x)$ is strongly convex and function $g(y)$ is affine, or both functions are affine, there exist lower bounds on the number of gradient evaluations and matrix-vector multiplications required to solve the problem, as well as matching optimal algorithms. A notable aspect of these algorithms is that they are able to attain linear convergence, i.e., the number of iterations required to solve the problem is proportional to $\log(1/\epsilon)$. However, the class of bilinearly-coupled saddle-point problems for which linear convergence is possible is much wider and can involve general smooth non-strongly convex functions $f(x)$ and $g(y)$. Therefore, *we develop the first lower complexity bounds and matching optimal linearly converging algorithms for this problem class*. Our lower complexity bounds are much more general, but they cover and unify the existing results in the literature. On the other hand, our algorithm implements the separation of complexities, which, for the first time, enables the simultaneous achievement of both optimal gradient evaluation and matrix-vector multiplication complexities, resulting in the best theoretical performance to date.

Paperid:327

Authors:Zecheng Tang · Zechen Sun · Juntao Li · Zhu Qiaoming · Min Zhang

Title: LOGO --- Long cOntext aliGnment via efficient preference Optimization

Abstract:
Longcontext models (LCMs) have shown great potential in processing long input sequences (even more than 100M tokens) conveniently and effectively. With significant progress, recent research has pointed out that LCMs can accurately locate token-level salient information within the context. Yet, the generation performance of these LCMs is far from satisfactory and might result in misaligned responses, such as hallucinations. To enhance the generation capability of LCMs, existing works have investigated the effects of data size and quality for both pre-training and instruction tuning. Though achieving meaningful improvement, previous methods fall short in either effectiveness or efficiency. In this paper, we introduce LOGO (Long cOntext aliGnment via efficient preference Optimization), a training strategy that first introduces preference optimization for long-context alignment. To overcome the GPU memory-bound issue caused by the long sequence, LOGO employs a reference-free preference optimization strategy and adopts a position synthesis method to construct the training data. By training with only 0.3B data on a single 8 x A800 GPU machine for 16 hours, LOGO allows the Llama-3-8B-Instruct-80K model to achieve comparable performance with GPT-4 in real-world long-context tasks while preserving the model's original capabilities on other tasks, e.g., language modeling and MMLU. Moreover, LOGO can extend the model's context window size while enhancing its generation performance.

Paperid:328

Authors:Zhiwei XU · Kun Hu · Xin Xin · Weiliang Meng · Yiwei Shi · Hangyu Mao · Bin Zhang · dapeng Li · Jiangjin Yin

Title: Reidentify: Context-Aware Identity Generation for Contextual Multi-Agent Reinforcement Learning

Abstract:
Generalizing multiagent reinforcement learning (MARL) to accommodate variations in problem configurations remains a critical challenge in real-world applications, where even subtle differences in task setups can cause pre-trained policies to fail. To address this, we propose Context-Aware Identity Generation (CAID), a novel framework to enhance MARL performance under the Contextual MARL (CMARL) setting. CAID dynamically generates unique agent identities through the agent identity decoder built on a causal Transformer architecture. These identities provide contextualized representations that align corresponding agents across similar problem variants, facilitating policy reuse and improving sample efficiency. Furthermore, the action regulator in CAID incorporates these agent identities into the action-value space, enabling seamless adaptation to varying contexts. Extensive experiments on CMARL benchmarks demonstrate that CAID significantly outperforms existing approaches by enhancing both sample efficiency and generalization across diverse context variants.

Paperid:329

Authors:Tianjie Ju · Yi Hua · Hao Fei · Zhenyu Shao · Yubin Zheng · Haodong Zhao · Mong-Li Lee · Wynne Hsu · Zhuosheng Zhang · Gongshen Liu

Title: Watch Out Your Album! On Unintentional Privacy Memorization in Multi-Modal Large Language Models

Abstract:
MultiModal Large Language Models (MLLMs) have exhibited remarkable performance on various vision-language tasks such as Visual Question Answering (VQA). Despite accumulating evidence of privacy concerns associated with task-relevant content, it remains unclear whether MLLMs inadvertently memorize private content that is entirely irrelevant to the training tasks. In this paper, we investigate how randomly generated task-irrelevant private content can become spuriously correlated with downstream objectives due to partial mini-batch training dynamics, thus causing inadvertent memorization. Concretely, we randomly generate task-irrelevant watermarks into VQA fine-tuning images at varying probabilities and propose a novel probing framework to determine whether MLLMs have inadvertently encoded such content. Our experiments reveal that MLLMs exhibit notably different training behaviors in partial mini-batch settings with task-irrelevant watermarks embedded. Furthermore, through layer-wise probing, we demonstrate that MLLMs trigger distinct representational patterns when encountering previously seen task-irrelevant knowledge, even if this knowledge does not influence their output during prompting. Our code is available atanonymity.

Paperid:330

Authors:Nadav Timor · Jonathan Mamou · Daniel Korat · Moshe Berchansky · Oren Pereg · Gaurav Jain · Moshe Wasserblat · David Harel

Title: Accelerating LLM Inference with Lossless Speculative Decoding for Heterogeneous Vocabularies

Abstract:
Accelerating the inference of large language models (LLMs) is a critical challenge in generative AI. Speculative decoding (SD) methods offer substantial efficiency gains by generating multiple tokens using a single target forward pass. However, existing SD approaches require the drafter and target models to share the same vocabulary, thus limiting the pool of possible drafters, often necessitating the training of a drafter from scratch. We present three new SD methods that remove this sharedvocabulary constraint. All three methods preserve the target distribution (i.e., they are lossless) and work with off-the-shelf models without requiring additional training or modifications. Empirically, on summarization and long-context tasks, our algorithms achieve significant speedups over standard autoregressive decoding. By enabling any off-the-shelf model to serve as drafter and requiring no retraining, this work substantially broadens the applicability of the SD framework in practice.

Paperid:331

Authors:Chenguang Wang · Kaiyuan Cui · Weichen Zhao · Tianshu Yu

Title: Sampling from Binary Quadratic Distributions via Stochastic Localization

Abstract:
Sampling from binary quadratic distributions (BQDs) is a fundamental but challenging problem in discrete optimization and probabilistic inference. Previous work established theoretical guarantees for stochastic localization (SL) in continuous domains, where MCMC methods efficiently estimate the required posterior expectations during SL iterations. However, achieving similar convergence guarantees for discrete MCMC samplers in posterior estimation presents unique theoretical challenges. In this work, we present the first application of SL to general BQDs, proving that after a certain number of iterations, the external field of posterior distributions constructed by SL tends to infinity almost everywhere, hence satisfy Poincaré inequalities with probability near to 1, leading to polynomialtime mixing. This theoretical breakthrough enables efficient sampling from general BQDs, even those that may not originally possess fast mixing properties. Furthermore, our analysis, covering enormous discrete MCMC samplers based on Glauber dynamics and Metropolis-Hastings algorithms, demonstrates the broad applicability of our theoretical framework.Experiments on instances with quadratic unconstrained binary objectives, including maximum independent set, maximum cut, and maximum clique problems, demonstrate consistent improvements in sampling efficiency across different discrete MCMC samplers.

Paperid:332

Authors:Yaolun Zhang · Xiaogeng Liu · Chaowei Xiao

Title: MetaAgent: Automatically Building Multi-Agent System based on Finite State Machine

Abstract:
Large Language Models (LLMs) can solve various practical tasks via a multiagent system. However, existing human-designed multi-agent systems can only adapt to a limited number of pre-defined scenarios. Current auto-designed methods also have several drawbacks, including no tool support, reliance on external training data, and inflexible communication structure. Therefore, we propose \textbf{MetaAgent}, a novel framework to automatically generate a multi-agent system based on a \textbf{finite state machine}. Given a task description, MetaAgent will design a multi-agent system and polish it through an optimization algorithm that merges redundant states. When the multi-agent system is deployed, the finite state machine will control the agent's actions and the state transitions. To evaluate our framework, we conduct experiments on both text-based tasks and practical tasks, the results indicate that the generated multi-agent system surpasses other auto-designed methods and can achieve a comparable performance with the human-designed multi-agent system which is polished for those specific tasks.

Paperid:333

Authors:Tianci Liu · Ruirui Li · Zihan Dong · Hui Liu · Xianfeng Tang · Qingyu Yin · Linjun Zhang · Haoyu Wang · Jing Gao

Title: Mitigating Heterogeneous Token Overfitting in LLM Knowledge Editing

Abstract:
Large language models (LLMs) have achieved remarkable performance on various natural language tasks. However, they are trained on static corpora and their knowledge can become outdated quickly in the fastchanging world. This motivates the development of knowledge editing (KE) to update specific knowledge in LLMs without changing unrelated others or compromising their pre-trained capabilities. Previous efforts sought to update a small amount of parameters of a LLM and proved effective for making selective updates. Nonetheless, the edited LLM often exhibits degraded ability to reason about the new knowledge. In this work, we identify a key issue: \textit{heterogeneous token overfitting} (HTO), where the LLM overfits different tokens in the provided knowledge at varying rates.To tackle this, we propose {\name}, a token-level smoothing method that mitigates HTO by adaptively refining the target distribution. Theoretically, OVERTONE offers better parameter updates with negligible computation overhead. It also induces an implicit DPO but does not require preference data pairs. Extensive experiments across four editing methods, two LLMs, and diverse scenarios demonstrate the effectiveness and versatility of our method.

Paperid:334

Authors:Yizi Zhang · Yanchen Wang · Mehdi Azabou · Alexandre Andre · Zixuan Wang · Hanrui Lyu · International Brain Laboratory · Eva Dyer · Department of Statistics Liam Paninski · Cole Hurwitz

Title: Neural Encoding and Decoding at Scale

Abstract:
Recent work has demonstrated that largescale, multi-animal models are powerful tools for characterizing the relationship between neural activity and behavior. Current large-scale approaches, however, focus exclusively on either predicting neural activity from behavior (encoding) or predicting behavior from neural activity (decoding), limiting their ability to capture the bidirectional relationship between neural activity and behavior. To bridge this gap, we introduce a multimodal, multi-task model that enables simultaneous Neural Encoding and Decoding at Scale (NEDS). Central to our approach is a novel multi-task-masking strategy, which alternates between neural, behavioral, within-modality, and cross-modality masking. We pretrain our method on the International Brain Laboratory (IBL) repeated site dataset, which includes recordings from 83 animals performing the visual decision-making task. In comparison to other large-scale modeling approaches, we demonstrate that NEDS achieves state-of-the-art performance for both encoding and decoding when pretrained on multi-animal data and then fine-tuned on new animals. Surprisingly, NEDS's learned embeddings exhibit emergent properties: even without explicit training, they are highly predictive of the brain regions in each recording. Altogether, our approach is a step towards a foundation model of the brain that enables seamless translation between neural activity and behavior.

Paperid:335

Authors:Ni Mu · Hao Hu · Xiao Hu · Yiqin Yang · Bo XU · Qing-Shan Jia

Title: CLARIFY: Contrastive Preference Reinforcement Learning for Untangling Ambiguous Queries

Abstract:
Preferencebased reinforcement learning (PbRL) bypasses explicit reward engineering by inferring reward functions from human preference comparisons, enabling better alignment with human intentions. However, humans often struggle to label a clear preference between similar segments, reducing label efficiency and limiting PbRL’s real-world applicability. To address this, we propose an offline PbRL method: Contrastive LeArning for ResolvIng Ambiguous Feedback (CLARIFY), which learns a trajectory embedding space that incorporates preference information, ensuring clearly distinguished segments are spaced apart, thus facilitating the selection of more unambiguous queries. Extensive experiments demonstrate that CLARIFY outperforms baselines in both non-ideal teachers and real human feedback settings. Our approach not only selects more distinguished queries but also learns meaningful trajectory embeddings.

Paperid:336

Authors:Chenyin Gao · Shu Yang · Mingyang Shan · Wenyu Ye · Ilya Lipkovich · Douglas Faries

Title: Doubly Protected Estimation for Survival Outcomes Utilizing External Controls for Randomized Clinical Trials

Abstract:
Censored survival data are common in clinical trials, but small control groups can pose challenges, particularly in rare diseases or where balanced randomization is impractical. Recent approaches leverage external controls from historical studies or realworld data to strengthen treatment evaluation for survival outcomes. However, using external controls directly may introduce biases due to data heterogeneity. We propose a doubly protected estimator for the treatment-specific restricted mean survival time difference that is more efficient than trial-only estimators and mitigates biases from external data. Our method adjusts for covariate shifts via doubly robust estimation and addresses outcome drift using the DR-Learner for selective borrowing. The approach incorporates machine learning to approximate survival curves and detect outcome drifts without strict parametric assumptions, borrowing only comparable external controls. Extensive simulation studies and a real-data application evaluating the efficacy of Galcanezumab in mitigating migraine headaches have been conducted to illustrate the effectiveness of our proposed framework.

Paperid:337

Authors:Yanxiang Ma · Zixuan Huang · Minjing Dong · Shan You · Chang Xu

Title: Adversarial Robustness via Deformable Convolution with Stochasticity

Abstract:
Random defense represents a promising strategy to protect neural networks from adversarial attacks. Most of these methods enhance robustness by injecting randomness into the data, increasing uncertainty for attackers.However, this randomness could reduce the generalization capacity of defense, as defense performance could be sensitive to the hyperparameters of noise added to the data, making it difficult to generalize across different datasets. Additionally, the involvement of randomness always comes with a reduction of natural accuracy, which leads to a delicate tradeoff between them, which is seldom studied in random defense. In this work, we propose incorporating randomness into the network structure instead of data input by designing stochastic deformable convolution, where a random mask replaces the convolutional offset. This process promotes data independence, enhancing generalization across datasets. To study the trade-off, we conduct a theoretical analysis of both robust and clean accuracy, from a perspective of gradient cosine similarity and natural inference. Based on the analysis, we reformulate the adversarial training in our random defense framework. Extensive experiments show that our method achieves SOTA adversarial robustness and clean accuracy compared with other random defense methods.

Paperid:338

Authors:Tao Tang · Lijun Zhou · Pengkun Hao · Zihang He · Kalok Ho · Shuo Gu · Zhihui Hao · Haiyang Sun · Kun Zhan · Peng Jia · XianPeng Lang · Xiaodan Liang

Title: S2-Track: A Simple yet Strong Approach for End-to-End 3D Multi-Object Tracking

Abstract:
3D multiple object tracking (MOT) plays a crucial role in autonomous driving perception. Recent endto-end query-based trackers simultaneously detect and track objects, which have shown promising potential for the 3D MOT task. However, existing methods are still in the early stages of development and lack systematic improvements, failing to track objects in certain complex scenarios, like occlusions and the small size of target object’s situations. In this paper, we first summarize the current end-to-end 3D MOT framework by decomposing it into three constituent parts: query initialization, query propagation, and query matching. Then we propose corresponding improvements, which lead to a strong yet simple tracker: S2-Track. Specifically, for query initialization, we present 2D-Prompted Query Initialization, which leverages predicted 2D object and depth information to prompt an initial estimate of the object’s 3D location. For query propagation, we introduce an Uncertainty-aware Probabilistic Decoder to capture the uncertainty of complex environment in object prediction with probabilistic attention. For query matching, we propose a Hierarchical Query Denoising strategy to enhance training robustness and convergence. As a result, our S2-Track achieves state-of-the-art performance on nuScenes benchmark, i.e., 66.3% AMOTA on test split, surpassing the previous best end-to-end solution by a significant margin of 8.9% AMOTA.

Paperid:339

Authors:Yafei YANG · Zihui Zhang · Bo Yang

Title: OCN: Learning Object-centric Representations for Unsupervised Multi-object Segmentation

Abstract:
We study the challenging problem of unsupervised multiobject segmentation on single images. Existing methods, which rely on image reconstruction objectives to learn objectness or leverage pretrained image features to group similar pixels, often succeed only in segmenting simple synthetic objects or discovering a limited number of real-world objects. In this paper, we introduce OCN, a novel two-stage pipeline designed to identify many complex objects in real-world images. The key to our approach involves explicitly learning three levels of carefully defined object-centric representations in the first stage. Subsequently, our multi-object reasoning module utilizes these learned object priors to discover multiple objects in the second stage. Notably, this reasoning module is entirely network-free and does not require human labels. Extensive experiments demonstrate that OCN significantly outperforms all existing unsupervised methods across 6 real-world benchmark datasets, including the challenging COCO dataset, achieving state-of-the-art object segmentation results. Remarkably, our method excels in crowded images where all baselines collapse.

Paperid:340

Authors:Zhengyu Zhou · Weiwei Liu

Title: An Error Analysis of Flow Matching for Deep Generative Modeling

Abstract:
Continuous Normalizing Flows (CNFs) have proven to be a highly efficient technique for generative modeling of complex data since the introduction of Flow Matching (FM). The core of FM is to learn the constructed velocity fields of CNFs through deep least squares regression. Despite its empirical effectiveness, theoretical investigations of FM remain limited. In this paper, we present the first endto-end error analysis of CNFs built upon FM. Our analysis shows that for general target distributions with bounded support, the generated distribution of FM is guaranteed to converge to the target distribution in the sense of the Wasserstein-2 distance. Furthermore, the convergence rate is significantly improved under an additional mild Lipschitz condition of the target score function.

Paperid:341

Authors:Mang Ning · Mingxiao Li · Jianlin Su · Jia Haozhe · Lanmiao Liu · Martin Benes · Wenshuo Chen · Albert Ali Salah · Itir Onal Ertugrul

Title: DCTdiff: Intriguing Properties of Image Generative Modeling in the DCT Space

Abstract:
Abstract:This paper explores image modeling from the frequency space and introduces DCTdiff, an endto-end diffusion generative paradigm that efficiently models images in the discrete cosine transform (DCT) space. We investigate the design space of DCTdiff and reveal the key design factors. Experiments on different frameworks (UViT, DiT), generation tasks, and various diffusion samplers demonstrate that DCTdiff outperforms pixel-based diffusion models regarding generative quality and training efficiency. Remarkably, DCTdiff can seamlessly scale up to 512$\times$512 resolution without using the latent diffusion paradigm and beats latent diffusion with only 1/4 training cost. Finally, we illustrate several intriguing properties of DCT image modeling. For example, we provide a theoretical proof of why `image diffusion can be seen as spectral autoregression', bridging the gap between diffusion and autoregressive models. The effectiveness of DCTdiff and the introduced properties suggest a promising direction for image modeling in the frequency space. The code is available.

Paperid:342

Authors:Peipeng Yu · Hui Gao · Xuan Feng · Jianwei Fei · Zhihua Xia · Chip Hong Chang

Title: Unlocking the Capabilities of Vision-Language Models for Generalizable Deepfake Detection

Abstract:
Current visionlanguage models (VLMs) have demonstrated remarkable capabilities in understanding multimodal data, but their potential remains underexplored for deepfake detection due to the misaligned of their knowledge and forensics patterns. To this end, we present a novel paradigm that unlocks VLMs' potential capabilities through three components: (1) A knowledge-guided forgery adaptation module that aligns VLM's semantic space with forensic features through contrastive learning with external manipulation knowledge; (2) A multi-modal prompt tuning framework that jointly optimizes visual-textual embeddings for both localization and explainability; (3) An iterative refinement strategy enabling multi-turn dialog for evidence-based reasoning. Our framework includes a VLM-based Knowledge-guided Forgery Detector (KFD), a VLM image encoder, and a Large Language Model (LLM). The VLM image encoder extracts visual prompt embeddings from images, while the LLM receives visual and question prompt embeddings for inference. The KFD is used to calculate correlations between image features and pristine/deepfake class embeddings, enabling forgery classification and localization. The outputs from these components are used to construct forgery prompt embeddings. Finally, we feed these prompt embeddings into the LLM to generate textual detection responses to assist judgment. Extensive experiments on multiple benchmarks, including FF++, CDF2, DFD, DFDCP, and DFDC, demonstrate that our scheme surpasses state-of-the-art methods in generalization performance, while also supporting multi-turn dialogue capabilities.

Paperid:343

Authors:yu chen · Jing Lian · Zhaofei Yu · Jizhao Liu · Jisheng Dang · Gang Wang

Title: A Chaotic Dynamics Framework Inspired by Dorsal Stream for Event Signal Processing

Abstract:
Event cameras are bioinspired vision sensors that encode visual information with high dynamic range, high temporal resolution, and low latency. Current state-of-the-art event stream processing methods rely on end-to-end deep learning techniques. However, these models are heavily dependent on data structures, limiting their stability and generalization capabilities across tasks, thereby hindering their deployment in real-world scenarios. To address this issue, we propose a chaotic dynamics event signal processing framework inspired by the dorsal visual pathway of the brain. Specifically, we utilize Continuous-coupled Neural Network (CCNN) to encode the event stream. CCNN encodes polarity-invariant event sequences as periodic signals and polarity-changing event sequences as chaotic signals. We then use continuous wavelet transforms to analyze the dynamical states of CCNN neurons and establish the high-order mappings of the event stream. The effectiveness of our method is validated through integration with conventional classification networks, achieving state-of-the-art classification accuracy on the N-Caltech101 and N-CARS datasets, with results of 84.3% and 99.9%, respectively. Our method improves the accuracy of event camera-based object classification while significantly enhancing the generalization and stability of event representation.

Paperid:344

Authors:Bing Liu · wenjun Miao · Boyu Zhang · Qiankun Zhang · Bin Yuan · Wang · Shenghao Liu · Xianjun Deng

Title: Understanding the Unfairness in Network Quantization

Abstract:
Network quantization, one of the most widely studied model compression methods, effectively quantizes a floatingpoint model to obtain a fixed-point one with negligible accuracy loss. Although great success was achieved in reducing the model size, it may exacerbate the unfairness in model accuracy across different groups of datasets.This paper considers two widely used algorithms: Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT), with an attempt to understand how they cause this critical issue.Theoretical analysis with empirical verifications reveals two responsible factors, as well as how they influence a metric of fairness in depth.A comparison between PTQ and QAT is then made, explaining an observation that QAT behaves even worse than PTQ in fairness, although it often preserves a higher accuracy at lower bit-widths in quantization.Finally, the paper finds out that several simple data augmentation methods can be adopted to alleviate the disparate impacts of quantization, based on a further observation that class imbalance produces distinct values of the aforementioned factors among different attribute classes. We experiment on either imbalanced (UTK-Face and FER2013) or balanced (CIFAR-10 and MNIST) datasets using ResNet and VGG models for empirical evaluation.

Paperid:345

Authors:Zhenyu Sun · Ermin Wei

Title: Improved Lower Bounds for First-order Stochastic Non-convex Optimization under Markov Sampling

Abstract:
Abstract:Unlike its vanilla counterpart with i.i.d. samples, stochastic optimization with Markovian sampling allows the sampling scheme following a Markov chain. This problem encompasses various applications that range from asynchronous distributed optimization to reinforcement learning. In this work, we lower bound the sample complexity of finding $\epsilon$approximate critical solutions for any first-order methods when sampling is Markovian. We show that for samples drawn from stationary Markov processes with countable state space, any algorithm that accesses smooth, non-convex functions through queries to a stochastic gradient oracle, requires at least $\Omega(\epsilon^{-4})$ samples. Moreover, for finite Markov chains, we show a $\Omega(\epsilon^{-2})$ lower bound and propose a new algorithm, called MaC-SAGE, that is proven to (nearly) match our lower bound.

Paperid:346

Authors:Thomas Y.L. Lin · Jerry Yao-Chieh Hu · Wan-Jiun Paul Chiou · Peter Lin

Title: Latent Variable Estimation in Bayesian Black-Litterman Models

Abstract:
Abstract:We propose a Bayesian reformulation of the BlackLitterman model for portfolio optimization without the need for subjective investor views and corresponding uncertainties $(q, \Omega)$ from humans.Classical Black-Litterman models treat $(q, \Omega)$ as user-provided factors but offer no rigorous mechanism for deriving these views, limiting their practicality. By recasting the Black-Litterman model as a Bayesian network, we incorporate asset-related features, treat $(q, \Omega)$ as latent variables, and estimate posterior distributions over both asset returns $r$ and their parameters $\theta$ directly from data. This approach unifies feature integration and parameter inference into a single framework, mitigating error propagation and ensuring coherent estimations. We explore two modes of feature integration: (1) features sharing latent parameters with returns and (2) features influencing investor views.Importantly,our formulation subsumes standard Black-Litterman formulas as special cases when $(q, \Omega)$ are given. Numerically, empirical tests on real-world datasets consistently outperform both the Markowitz and market-index baselines without heuristic views.

Paperid:347

Authors:Harry Zhang · Luca Carlone

Title: CUPS: Improving Human Pose-Shape Estimators with Conformalized Deep Uncertainty

Abstract:
We introduce CUPS, a novel method for learning sequenceto-sequence 3D human shapes and poses from RGB videos with uncertainty quantification. To improve on top of prior work, we develop a method to {generate and score multiple hypotheses during training}, effectively integrating uncertainty quantification into the learning process. This process results in a deep uncertainty function that is trained end-to-end with the 3D pose estimator. Post-training, the learned deep uncertainty model is used as the conformity score, which can be used to calibrate a conformal predictor in order to {assess} the quality of the output prediction. Since the data in human pose-shape learning is not fully exchangeable, we also {present} two practical bounds for the coverage gap in conformal prediction, developing theoretical backing for the uncertainty bound of our model. Our results indicate thatby taking advantage of deep uncertainty with conformal prediction, our method achieves state-of-the-art performance across variousmetrics and datasets while inheriting the probabilistic guarantees of conformal prediction. Interactive 3D visualization, code, and data will be available at https://sites.google.com/view/champpp.

Paperid:348

Authors:Anshuman Chhabra · Bo Li · Jian Chen · Prasant Mohapatra · Hongfu Liu

Title: Outlier Gradient Analysis: Efficiently Identifying Detrimental Training Samples for Deep Learning Models

Abstract:
A core datacentric learning challenge is the identification of training samples that are detrimental to model performance. Influence functions serve as a prominent tool for this task and offer a robust framework for assessing training data influence on model predictions. Despite their widespread use, their high computational cost associated with calculating the inverse of the Hessian matrix pose constraints, particularly when analyzing large-sized deep models. In this paper, we establish a bridge between identifying detrimental training samples via influence functions and outlier gradient detection. This transformation not only presents a straightforward and Hessian-free formulation but also provides insights into the role of the gradient in sample impact. Through systematic empirical evaluations, we first validate the hypothesis of our proposed outlier gradient analysis approach on synthetic datasets. We then demonstrate its effectiveness in detecting mislabeled samples in vision models and selecting data samples for improving performance of natural language processing transformer models. We also extend its use to influential sample identification for fine-tuning Large Language Models.

Paperid:349

Authors:Mingyi Li · Michael Metel · Akiko Takeda

Title: Modified K-means Method with Local Optimality Guarantees

Abstract:
The Kmeans algorithm is one of the most widely studied clustering algorithms in machine learning. While extensive research has focused on trying to achieve global optima, no algorithm has been developed with even a rigorous local optimality guarantee. In this paper, we first present conditions under which the K-means algorithm converges to a locally optimal solution. Based on this, we propose simple modifications to the K-means algorithm which ensure local optimality in the continuous and discrete senses, with the same computational complexity as the ordinary K-means algorithm. As the dissimilarity measure, we consider a general Bregman divergence, which is an extension of the squared Euclidean distance often used in the K-means algorithm. Numerical experiments on synthetic data confirm that the ordinary K-means algorithm does not always find local minima in practice, while our proposed method provides improved locally optimal solutions with smaller objective function values.

Paperid:350

Authors:Ze'ev Zukerman · Bassel Hamoud · Kfir Levy

Title: Beyond Communication Overhead: A Multilevel Monte Carlo Approach for Mitigating Compression Bias in Distributed Learning

Abstract:
Abstract:Distributed learning methods have gained substantial momentum in recent years, with communication overhead often emerging as a critical bottleneck. Gradient compression techniques alleviate communication costs but involve an inherent tradeoff between the empirical efficiency of biased compressors and the theoretical guarantees of unbiased compressors. In this work, we introduce a novel Multilevel Monte Carlo (MLMC) compression scheme that leverages biased compressors to construct statistically unbiased estimates. This approach effectively bridges the gap between biased and unbiased methods, combining the strengths of both. To showcase the versatility of our method, we apply it to top-$k$ and bit-wise compressors, resulting in enhanced variants. Furthermore, we derive an adaptive version of our approach to further improve its performance. We validate our method empirically on distributed learning tasks.

Paperid:351

Authors:Zhangyi Liu · Feng Liu · Rui Gao · Shuang Li

Title: Convergence of Mean-Field Langevin Stochastic Descent-Ascent for Distributional Minimax Optimization

Abstract:
Abstract:We study convergence properties of the discretetime Mean-Field Langevin Stochastic Descent-Ascent (MFL-SDA) algorithm for solving distributional minimax optimization. These problems arise in various applications, such as zero-sum games, generative adversarial networks and distributionally robust learning. Despite the significance of MFL-SDA in these contexts, the discrete-time convergence rate remains underexplored.To address this gap, we establish a last-iterate convergence rate of $O(\frac{1}{\epsilon}\log\frac{1}{\epsilon})$ for MFL-SDA. This rate is nearly optimal when compared to the complexity lower bound of its Euclidean counterpart. This rate also matches the complexity of mean-field Langevin stochastic gradient descent for distributional minimization and the outer-loop iteration complexity of an existing double-loop algorithm for distributional minimax problems.By leveraging an elementary analysis framework that avoids PDE-based techniques, we overcome previous limitations and achieve a faster convergence rate.

Paperid:352

Authors:Zitao Wang · Ziyuan Wang · Molei Liu · Nian Si

Title: Knowledge-Guided Wasserstein Distributionally Robust Optimization

Abstract:
Transfer learning is a popular strategy to leverage external knowledge and improve statistical efficiency, particularly with a limited target sample. We propose a novel knowledgeguided Wasserstein Distributionally Robust Optimization (KG-WDRO) framework that incorporates multiple sources of external knowledge to overcome the conservativeness of traditional WDRO, which often results in overly pessimistic shrinkage toward zero. Our method constructs smaller Wasserstein ambiguity sets by controlling the transportation along directions informed by the source knowledge. This strategy can alleviate perturbations on the predictive projection of the covariates and protect against information loss. Theoretically, we establish the equivalence between our WDRO formulation and the knowledge-guided shrinkage estimation based on collinear similarity, ensuring tractability and geometrizing the feasible set. This also reveals a novel and general interpretation for recent shrinkage-based transfer learning approaches from the perspective of distributional robustness. In addition, our framework can adjust for scaling differences in the regression models between the source and target and accommodates general types of regularization such as lasso and ridge. Extensive simulations demonstrate the superior performance and adaptivity of KG-WDRO in enhancing small-sample transfer learning.

Paperid:353

Authors:Min Zhao · Guande He · Yixiao Chen · Hongzhou Zhu · Chongxuan Li · Jun Zhu

Title: A Free Lunch for Length Extrapolation in Video Diffusion Transformers

Abstract:
Abstract:Recent advancements in video generation have enabled models to synthesize highquality, minute-long videos. However, generating even longer videos with temporal coherence remains a major challenge and existing length extrapolation methods lead to \textit{temporal repetition} or \textit{motion deceleration}. In this work, we systematically analyze the role of frequency components in positional embeddings and identify an \textit{intrinsic frequency} that primarily governs extrapolation behavior. Based on this insight, we propose \textbf{RIFLEx}, a \textit{minimal yet effective} approach that reduces the intrinsic frequency to suppress repetition while preserving motion consistency, without requiring any additional modifications. RIFLEx offers a true \textit{free lunch}—achieving high-quality $2\times$ extrapolation on state-of-the-art video diffusion transformers in a completely \textit{training-free} manner. Moreover, it enhances quality and enables $3\times$ extrapolation by minimal fine-tuning without long videos.

Paperid:354

Authors:Hui Dai · Ryan Teehan · Mengye Ren

Title: Are LLMs Prescient? A Continuous Evaluation using Daily News as the Oracle

Abstract:
Many existing evaluation benchmarks for Large Language Models (LLMs) quickly become outdated due to the emergence of new models and training data. These benchmarks also fall short in assessing how LLM performance changes over time, as they consist of a static set of questions without a temporal dimension. To address these limitations, we propose using future event prediction as a continuous evaluation method to assess LLMs' temporal generalization and forecasting abilities. Our benchmark, Daily Oracle, automatically generates questionanswer (QA) pairs from daily news, challenging LLMs to predict "future" event outcomes. Our findings reveal that as pre-training data becomes outdated, LLM performance degrades over time. While Retrieval Augmented Generation (RAG) has the potential to enhance prediction accuracy, the performance degradation pattern persists, highlighting the need for continuous model updates.

Paperid:355

Authors:Andrea Testa · Søren Hauberg · Tamim Asfour · Leonel Rozo

Title: Geometric Contact Flows: Contactomorphisms for Dynamics and Control

Abstract:
Accurately modeling and predicting complex dynamical systems, particularly those involving contact and interaction, is crucial for applications ranging from fluid dynamics to robotics, but presents significant challenges due to the intricate interplay of geometric constraints and energy exchange. This paper introduces Geometric Contact Flows (GFC), a novel framework leveraging Riemannian and Contact geometry as inductive biases to learn such systems. GCF constructs a latent contact Hamiltonian model encoding desirable properties like stability or energy conservation. An ensemble of contactomorphisms then adapts this model to the target dynamics, preserving these properties while enabling robust generalization and adaptation to unseen scenarios. This ensemble allows for uncertaintyaware geodesics that attract the system's behavior toward the data support.Experiments on handwriting dynamics tasks and a robotic interaction task demonstrate the effectiveness of our approach.

Paperid:356

Authors:Daoyu Wang · Mingyue Cheng · Zhiding Liu · Qi Liu

Title: TimeDART: A Diffusion Autoregressive Transformer for Self-Supervised Time Series Representation

Abstract:
Selfsupervised learning has garnered increasing attention in time series analysis for benefiting various downstream tasks and reducing reliance on labeled data. Despite its effectiveness, existing methods often struggle to comprehensively capture both long-term dynamic evolution and subtle local patterns in a unified manner. In this work, we propose \textbf{TimeDART}, a novel self-supervised time series pre-training framework that unifies two powerful generative paradigms to learn more transferable representations. Specifically, we first employ a \textit{causal} Transformer encoder, accompanied by a patch-based embedding strategy, to model the evolving trends from left to right. Building on this global modeling, we further introduce a denoising diffusion process to capture fine-grained local patterns through forward diffusion and reverse denoising. Finally, we optimize the model in an autoregressive manner. As a result, TimeDART effectively accounts for both global and local sequence features in a coherent way. We conduct extensive experiments on public datasets for time series forecasting and classification. The experimental results demonstrate that TimeDART consistently outperforms previous compared methods, validating the effectiveness of our approach. Our code is available at \url{https://anonymous.4open.science/r/TimeDART-2025/}.

Paperid:357

Authors:Yunzhuo Hao · Jiawei Gu · Huichen Wang · Linjie Li · Zhengyuan Yang · Lijuan Wang · Yu Cheng

Title: Can MLLMs Reason in Multimodality? EMMA: An Enhanced MultiModal ReAsoning Benchmark

Abstract:
The ability to organically reason over and with both text and images is a pillar of human intelligence, yet the ability of Multimodal Large Language Models (MLLMs) to perform such multimodal reasoning remains underexplored. Existing benchmarks often emphasize text-dominant reasoning or rely on shallow visual cues, failing to adequately assess integrated visual and textual reasoning. We introduce EMMA (Enhanced MultiModal reAsoning), a benchmark targeting organic multimodal reasoning across mathematics, physics, chemistry, and coding. EMMA tasks demand advanced cross-modal reasoning that cannot be addressed by reasoning independently in each modality, offering an enhanced test suite for MLLMs' reasoning capabilities. Our evaluation of state-of-the-art MLLMs on EMMA reveals significant limitations in handling complex multimodal and multi-step reasoning tasks, even with advanced techniques like Chain-of-Thought prompting and test-time compute scaling underperforming. These findings underscore the need for improved multimodal architectures and training paradigms to close the gap between human and model reasoning in multimodality.

Paperid:358

Authors:Jiaqi Leng · Bin Shi

Title: Quantum Optimization via Gradient-Based Hamiltonian Descent

Abstract:
With rapid advancements in machine learning, firstorder algorithms have emerged as the backbone of modern optimization techniques, owing to their computational efficiency and low memory requirements. Recently, the connection between accelerated gradient methods and damped heavy-ball motion, particularly within the framework of Hamiltonian dynamics, has inspired the development of innovative quantum algorithms for continuous optimization. One such algorithm, Quantum Hamiltonian Descent (QHD), leverages quantum tunneling to escape saddle points and local minima, facilitating the discovery of global solutions in complex optimization landscapes. However, QHD faces several challenges, including slower convergence rates compared to classical gradient methods and limited robustness in highly non-convex problems due to the non-local nature of quantum states. Furthermore, the original QHD formulation primarily relies on function value information, which limits its effectiveness. Inspired by insights from high-resolution differential equations that have elucidated the acceleration mechanisms in classical methods, we propose an enhancement to QHD by incorporating gradient information, leading to what we call gradient-based QHD. This gradient-based QHD achieves faster convergence and significantly increases the likelihood of identifying global solutions. Numerical simulations on challenging problem instances demonstrate that this gradient-based QHD outperforms existing quantum and classical methods by at least an order of magnitude.

Paperid:359

Authors:Yuqin Cao · Xiongkuo Min · Yixuan Gao · Wei Sun · Guangtao Zhai

Title: AGAV-Rater: Adapting Large Multimodal Model for AI-Generated Audio-Visual Quality Assessment

Abstract:
Many videoto-audio (VTA) methods have been proposed for dubbing silent AI-generated videos. An efficient quality assessment method for AI-generated audio-visual content (AGAV) is crucial for ensuring audio-visual quality. Existing audio-visual quality assessment methods struggle with unique distortions in AGAVs, such as unrealistic and inconsistent elements. To address this, we introduceAGAVQA, the first large-scale AGAV quality assessment dataset, comprising 3,382 AGAVs from 16 VTA methods. AGAVQA includes two subsets: AGAVQA-MOS, which provides multi-dimensional scores for audio quality, content consistency, and overall quality, and AGAVQA-Pair, designed for optimal AGAV pair selection. We further proposeAGAV-Rater, a LMM-based model that can score AGAVs, as well as audio and music generated from text, across multiple dimensions, and selects the best AGAV generated by VTA methods to present to the user. AGAV-Rater achieves state-of-the-art performance on AGAVQA, Text-to-Audio, and Text-to-Music datasets. Subjective tests also confirm that AGAV-Rater enhances VTA performance and user experience.

Paperid:360

Authors:Yudong Hu · Haoran Li · Congying Han · Tiande Guo · Bonan Li · Mingqiang Li

Title: A-PSRO: A Unified Strategy Learning Method with Advantage Metric for Normal-form Games

Abstract:
Solving the Nash equilibrium in normalform games with large-scale strategy spaces presents significant challenges. Open-ended learning frameworks, such as PSRO and its variants, have emerged as effective solutions. However, these methods often lack an efficient metric for evaluating strategy improvement, which limits their effectiveness in approximating equilibria.In this paper, we introduce a novel evaluative metric called Advantage, which possesses desirable properties inherently connected to the Nash equilibrium, ensuring that each strategy update approaches equilibrium. Building upon this, we propose the Advantage Policy Space Response Oracle~(A-PSRO), an innovative unified open-ended learning framework applicable to both zero-sum and general-sum games. A-PSRO leverages the Advantage as a refined evaluation metric, leading to a consistent learning objective for agents in normal-form games. Experiments showcase that A-PSRO significantly reduces exploitability in zero-sum games and improves rewards in general-sum games, outperforming existing algorithms and validating its practical effectiveness.

Paperid:361

Authors:Honghua Zhang · Meihua Dang · Benjie Wang · Stefano Ermon · Nanyun Peng · Guy Van den Broeck

Title: Scaling Probabilistic Circuits via Monarch Matrices

Abstract:
Probabilistic Circuits (PCs) are tractable representations of probability distributions allowing for exact and efficient computation of likelihoods and marginals. Recent advancements have improved the scalability of PCs by leveraging their sparse properties or tensorized operations. However, no existing method fully exploits both aspects simultaneously. In this paper, we propose a novel sparse and structured parameterization for the sum blocks in PCs. By replacing dense matrices with sparse Monarch matrices, we significantly reduce the memory and computation costs, scaling PCs to next level. From a theory perspective, our construction arises naturally from circuit multiplication; from a practical perspective, compared to previous efforts on scaling up tractable probabilistic models, our approach not only achieves stateof-the-art generative modeling performance on challenging benchmarks like Text8, LM1B and ImageNet, but also demonstrates superior scaling behavior, achieving the same performance with substantially less compute as measured by the number of floating-point operations (FLOPs) during training.

Paperid:362

Authors:Matthew Faw · Rajat Sen · Yichen Zhou · Abhimanyu Das

Title: In-Context Fine-Tuning for Time-Series Foundation Models

Abstract:
Motivated by the recent success of timeseries foundation models for zero-shot forecasting, we present a methodology forin-context fine-tuningof a time-series foundation model. In particular, we design a pretrained foundation model that can be prompted (at inference time) with multiple time-series examples, in order to forecast a target time-series into the future. Our foundation model is specifically trained to utilize examples from multiple related time-series in its context window (in addition to the history of the target time-series) to help it adapt to the specific distribution of the target domain at inference time. We show that such a foundation model that uses in-context examples at inference time can obtain much better performance on popular forecasting benchmarks compared to supervised deep learning methods, statistical models, and other time series foundation models. Interestingly, our in-context fine-tuning approach even matches the performance of a foundation model that is explicitly fine-tuned on the target domain.

Paperid:363

Authors:Suchith Prabhu · Bhavyajeet Singh · Anshul Mittal · Siddarth Asokan · Shikhar Mohan · Deepak Saini · Yashoteja Prabhu · Lakshya Kumar · Jian Jiao · Amit Singh · Niket Tandon · Manish Gupta · Sumeet Agarwal · Manik Varma

Title: MOGIC: METADATA-INFUSED ORACLE GUIDANCE FOR IMPROVED EXTREME CLASSIFICATION

Abstract:
Retrievalaugmented classification and generation models significantly benefit from theearly-stage fusionof high-quality text-based auxiliary metadata, often called memory, but they suffer from high inference latency and poor robustness to noise. In classifications tasks, particularly the extreme classification (XC) setting, where low latency is critical, existing methods incorporate metadata for context enrichment via an XC-based retriever and obtain the representations of the relevant memory items to performlate-stagefusion to achieve low latency and improved resilience to noisy metadata. With an aim of achieving higher accuracy while meeting the low latency constraints, in this paper, we propose MOGIC, an approach for metadata-infused Oracle guidance for XC tasks. In particular, we train an early-fusion Oracle classifier with access to both query-side and label-side ground-truth metadata in the textual form. The Oracle is subsequently used to guide the training of any existing memory-based XC disciple model via regularization. The MOGIC algorithm, when applied to memory-based XC disciple models such as OAK, improves precision@1 and propensity-scored precision@1 by ∼2% on six standard datasets, at no additional inference time costs to the disciple model. We also show the feasibility of applying the MOGIC algorithm to improve the performance of state-of-the-art memory-free XC approaches such as NGAME or DEXA, demonstrating that the MOGIC algorithm can be used atop any existing XC-based approach in aplug-and-playmanner. Finally, we also show the robustness of the MOGIC method to missing and noisy metadata settings. We will release code on acceptance.

Paperid:364

Authors:Shiyu Wu · Jing Liu · Jing Li · Yequan Wang

Title: Few-Shot Learner Generalizes Across AI-Generated Image Detection

Abstract:
Abstract:Current fake image detectors trained on large synthetic image datasets perform satisfactorily on limited studied generative models. However, these detectors suffer a notable performance decline over unseen models. Besides, collecting adequate training data from online generative models is often expensive or infeasible. To overcome these issues, we propose FewShot Detector (FSD), a novel AI-generated image detector which learns a specialized metric space to effectively distinguish unseen fake images by utilizing very few samples. Experiments show that FSD achieves state-of-the-art performance by $+7.4\%$ average ACC on GenImage dataset. More importantly, our method is better capable of capturing the intra-category common features in unseen images without further training.

Paperid:365

Authors:Yingying Feng · Jie Li · Chi Xie · Lei Tan · Jiayi Ji

Title: Multi-Modal Object Re-identification via Sparse Mixture-of-Experts

Abstract:
We present MFRNet, a novel network for multimodal object re-identification that integrates multi-spectral data features to effectively retrieve specific people across different modalities. Current methods suffer from two principal limitations: (1) insufficient interaction between pixel-level semantic features across modalities, and (2) difficulty in balancing modality-shared and modality-specific features within a unified architecture. To address these challenges, our network introduces two core components. First, the Feature Fusion Module (FFM) enables fine-grained pixel-level feature generation and flexible cross-modal interaction. Second, the Feature Representation Module (FRM) efficiently extracts and combines modality-specific and modality-shared features, achieving strong discriminative ability with minimal parameter overhead. Extensive experiments on three challenging public datasets (RGBNT201, RGBNT100, and MSVR310) demonstrate the superiority of our approach in terms of both accuracy and efficiency, with 8.4% mAP and 6.9% accuracy improved in RGBNT201 with nearly no extra computation.

Paperid:366

Authors:Arhit Chakrabarti · Yang Ni · Debdeep Pati · Bani Mallick

Title: Global-Local Dirichlet Processes for Clustering Grouped Data in the Presence of Group-Specific Idiosyncratic Variables

Abstract:
We consider the problem of clustering grouped data for which the observations may include groupspecific variables in addition to the variables that are shared across groups. This type of data is quite common; for example, in cancer genomic studies, molecular information is available for all cancers whereas cancer-specific clinical information may only be available for certain cancers. Existing grouped clustering methods only consider the shared variables but ignore valuable information from the group-specific variables. To allow for these group-specific variables to aid in the clustering, we propose a novel Bayesian nonparametric approach, termed global-local (GLocal) Dirichlet process, that models the ``global-local" structure of the observations across groups. We characterize the GLocal Dirichlet process using the stick-breaking representation and the representation as a limit of a finite mixture model. We theoretically quantify the approximation errors of the truncated prior, the corresponding finite mixture model, and the associated posterior distribution. We develop a fast variational Bayes algorithm for scalable posterior inference, which we illustrate with extensive simulations and a TCGA pan-gastrointestinal cancer dataset.

Paperid:367

Authors:Jason Chan · Robert Gaizauskas · Zhixue Zhao

Title: RULEBREAKERS: Challenging Large Language Models at the Crossroads between Formal Logic and Human-like Reasoning

Abstract:
Formal logic enables computers to reason in natural language by representing sentences in symbolic forms and applying rules to derive conclusions. However, in what our study characterizes as "rulebreaker" scenarios, this method can lead to conclusions that are typically not inferred or accepted by humans given their common sense and factual knowledge. Inspired by works in cognitive science, we create RULEBREAKERS, the first dataset for rigorously evaluating the ability of large language models (LLMs) to recognize and respond to rulebreakers (versus nonrulebreakers) in a human-like manner. Evaluating seven LLMs, we find that most models, including GPT-4o, achieve mediocre accuracy on RULEBREAKERS and exhibit some tendency to over-rigidly apply logical rules unlike what is expected from typical human reasoners. Further analysis suggests that this apparent failure is potentially associated with the models' poor utilization of their world knowledge and their attention distribution patterns. Whilst revealing a limitation of current LLMs, our study also provides a timely counterbalance to a growing body of recent works that propose methods relying on formal logic to improve LLMs' general reasoning capabilities, highlighting their risk of further increasing divergence between LLMs and human-like reasoning.

Paperid:368

Authors:Kaizhen Zhu · Mokai Pan · Yuexin Ma · Yanwei Fu · Jingyi Yu · Jingya Wang · Ye Shi

Title: UniDB: A Unified Diffusion Bridge Framework via Stochastic Optimal Control

Abstract:
Abstract:Recent advances in diffusion bridge models leverage Doob’s $h$transform to establish fixed endpoints between distributions, demonstrating promising results in image translation and restoration tasks. However, these approaches frequently produce blurred or excessively smoothed image details and lack a comprehensive theoretical foundation to explain these shortcomings. To address these limitations, we propose UniDB, a unified framework for diffusion bridges based on Stochastic Optimal Control (SOC). UniDB formulates the problem through an SOC-based optimization and derives a closed-form solution for the optimal controller, thereby unifying and generalizing existing diffusion bridge models. We demonstrate that existing diffusion bridges employing Doob’s $h$-transform constitute a special case of our framework, emerging when the terminal penalty coefficient in the SOC cost function tends to infinity. By incorporating a tunable terminal penalty coefficient, UniDB achieves an optimal balance between control costs and terminal penalties, substantially improving detail preservation and output quality. Notably, UniDB seamlessly integrates with existing diffusion bridge models, requiring only minimal code modifications. Extensive experiments across diverse image restoration tasks validate the superiority and adaptability of the proposed framework. Our code is available at https://github.com/UniDB-SOC/UniDB/.

Paperid:369

Authors:Jiayu Liu · Zhenya Huang · Chaokun Wang · Xunpeng Huang · Chengxiang Zhai · Enhong Chen

Title: What Makes In-context Learning Effective for Mathematical Reasoning: A Theoretical Analysis

Abstract:
Owing to the capability of incontext learning, large language models (LLMs) have shown impressive performance across diverse mathematical reasoning benchmarks. However, we find that few-shot demonstrations can sometimes bring negative performance and their effectiveness on LLMs' reasoning abilities remains unreliable. To this end, in this paper, we aim to theoretically analyze the impact of in-context demonstrations on LLMs' reasoning performance. We prove that the reasoning efficacy (measured by empirical prediction loss) can be bounded by a \emph{LLM-oriented semantic similarity} and an \emph{inference stability of demonstrations}, which is general for both one-shot and few-shot scenarios. Based on this finding, we propose a straightforward, generalizable, and low-complexity demonstration selection method named LMS3. It facilitates to select the most pertinent samples for different LLMs and includes a novel demonstration rejection mechanism to automatically filter out samples that are unsuitable for few-shot learning. Through experiments on three representative benchmarks, two LLM backbones, and multiple few-shot settings, we verify that our LMS3 has superiority and achieves consistent improvements on all datasets, which existing methods have been unable to accomplish.

Paperid:370

Authors:Gramoz Goranci · Peter Kiss · Neel Patel · Martin Seybold · Eva Szilagyi · Da Wei Zheng

Title: Fully Dynamic Euclidean Bi-Chromatic Matching in Sublinear Update Time

Abstract:
Abstract:We consider the Euclidean bichromatic matching problem in the dynamic setting, where the goal is to efficiently process point insertions and deletions while maintaining a high-quality solution. Computing the minimum cost bi-chromatic matching is one of the core problems in geometric optimization that has found many applications, most notably in estimating Wasserstein distance between two distributions. In this work, we present the first fully dynamic algorithm for Euclidean bi-chromatic matching with sublinear update time. For any fixed $\varepsilon > 0$, our algorithm achieves $O(1/\varepsilon)$-approximation and handles updates in $O(n^{\varepsilon})$ time. Our experiments show that our algorithm enables effective monitoring of the distributional drift in the Wasserstein distance on real and synthetic data sets, while outperforming the runtime of baseline approximations by orders of magnitudes.

Paperid:371

Authors:Wenxu Wang · Rui Zhou · Jing Wang · Bo Han · Yun Zhou · Cheng Zhu · Ruichun Tang · Nevin Zhang

Title: COSDA: Counterfactual-based Susceptibility Risk Framework for Open-Set Domain Adaptation

Abstract:
OpenSet Domain Adaptation (OSDA) aims to transfer knowledge from the labeled source domain to the unlabeled target domain that contains unknown categories, thus facing the challenges of domain shift and unknown category recognition. While recent works have demonstrated the potential of causality for domain alignment, little exploration has been conducted on causal-inspired theoretical frameworks for OSDA. To fill this gap, we introduce the concept ofSusceptibilityand propose a novelCounterfactual-based susceptibility risk framework forOSDA, termedCOSDA. Specifically, COSDA consists of three novel components: (i) aSusceptibility Risk Estimator (SRE)for capturing causal information, along with comprehensive derivations of the computable theoretical upper bound, forming a risk minimization framework under the OSDA paradigm; (ii) aContrastive Feature Alignment (CFA)module, which is theoretically proven based on mutual information to satisfy theExogeneityassumption and facilitate cross-domain feature alignment; (iii) aVirtual Multi-unknown-categories Prototype (VMP)pseudo-labeling strategy, providing label information by measuring how similar samples are to known and multiple virtual unknown category prototypes, thereby assisting in open-set recognition and intra-class discriminative feature learning. Extensive experiments on both benchmark and synthetic datasets demonstrate that our approach achieves state-of-the-art performance.

Paperid:372

Authors:Runyu Lu · Yuanheng Zhu · Dongbin Zhao

Title: Constrained Exploitability Descent: An Offline Reinforcement Learning Method for Finding Mixed-Strategy Nash Equilibrium

Abstract:
This paper proposes Constrained Exploitability Descent (CED), a modelfree offline reinforcement learning (RL) algorithm for solving adversarial Markov games. CED combines the game-theoretical approach of Exploitability Descent (ED) with policy constraint methods from offline RL. While policy constraints can perturb the optimal pure-strategy solutions in single-agent scenarios, we find the side effect less detrimental in adversarial games, where the optimal policy can be a mixed-strategy Nash equilibrium. We theoretically prove that, under the uniform coverage assumption on the dataset, CED converges to a stationary point in deterministic two-player zero-sum Markov games (MGs). We further prove that the min-player policy at the stationary point follows the property of mixed-strategy Nash equilibrium in MGs. Compared to the model-based ED method that optimizes the max-player policy, our CED method no longer relies on a generalized gradient. Experiments in matrix games, a tree-form game, and an infinite-horizon soccer game verify that CED can find an equilibrium policy for the min-player as long as the offline dataset guarantees uniform coverage. Besides, CED achieves a significantly lower NashConv compared to an existing pessimism-based method and can gradually improve the behavior policy even under non-uniform data coverages.

Paperid:373

Authors:Xiaojie Wang · Bin Yang

Title: Optimal Error Bounds in $\mathcal{W}_2$-Distance with Sqrt(d) Dimension Dependence for Langevin Monte Carlo beyond Log-Concavity

Abstract:
Abstract:Generating samples from a high dimensional probability distribution is a fundamental task with wideranging applications in the area of scientific computing, statistics and machine learning. This article revisits the popular Langevin Monte Carlo (LMC) sampling algorithms and provides an optimal non-asymptotic error analysis in $\mathcal{W}_2$-distance in a non-convex setting. In particular, we prove an optimal error bound $O(\sqrt{d} h)$, which guarantees a mixing time $ \tilde{O} (\sqrt{d} \epsilon^{-1})$ to achieve the accuracy tolerance $\epsilon$, under certain log-smooth conditions and the assumption that the target distribution satisfies a log-Sobolev inequality, as opposed to the strongly log-concave condition used in (Li et al., 2019; 2022). This bound matches the best one in the strongly log-concave case and improves upon the best-known convergence rates in non-convex settings. To prove it, we establish a new framework of uniform-in-time convergence for discretizations of SDEs. Distinct from (Li et al., 2019; 2022), we start from the finite-time mean-square fundamental convergence theorem, which combined with uniform-in-time moment bounds of LMC and the exponential ergodicity of SDEs in the non-convex setting gives the desired uniform-in-time convergence. Our framework also applies to the case when the gradient of the potential $U$ is non-globally Lipschitz with superlinear growth, for which modified LMC samplers are proposed and analyzed, with a non-asymptotic error bound in $\mathcal{W}_2$-distance obtained. Numerical experiments corroborate the theoretical analysis.

Paperid:374

Authors:Tobias Braun · Mark Rothermel · Marcus Rohrbach · Anna Rohrbach

Title: DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts

Abstract:
The proliferation of disinformation demands reliable and scalable factchecking solutions. We present Dynamic Evidence-based FAct-checking with Multimodal Experts (DEFAME), a modular, zero-shot MLLM pipeline for open-domain, text-image claim verification. DEFAME operates in a six-stage process, dynamically selecting the tools and search depth to extract and evaluate textual and visual evidence. Unlike prior approaches that are text-only, lack explainability, or rely solely on parametric knowledge, DEFAME performs end-to-end verification, accounting for images in claims and evidence while generating structured, multimodal reports. Evaluation on the popular benchmarks VERITE, AVerITeC, and MOCHEG shows that DEFAME surpasses all previous methods, establishing itself as the new state-of-the-art fact-checking system for uni- and multimodal fact-checking. Moreover, we introduce a new benchmark, CLAIMREVIEW24+, featuring claims after the knowledge cutoff of GPT4o to avoid data leakage. Here, DEFAME drastically outperforms the GPT Chain-of-Thought baseline, demonstrating temporal generalizability and the potential for real-time fact-checking.

Paperid:375

Authors:Amber Yijia Zheng · Cedar Site Bai · Brian Bullins · Raymond A. Yeh

Title: Model Immunization from a Condition Number Perspective

Abstract:
Model immunization aims to pretrain models that are difficult to fine-tune on harmful tasks while retaining their utility on other non-harmful tasks. Though prior work has shown empirical evidence for immunizing text-to-image models, the key understanding of when immunization is possible and a precise definition of an immunized model remains unclear. In this work, we propose a framework, based on the condition number of a Hessian matrix, to analyze model immunization for linear models. Building on this framework, we design an algorithm with regularization terms to control the resulting condition numbers after pre-training. Empirical results on linear models and non-linear deep-nets demonstrate the effectiveness of the proposed algorithm on model immunization.

Paperid:376

Authors:Wenbin Wang · Yu Shi · Ziping Zhao

Title: High-Dimensional Tensor Regression With Oracle Properties

Abstract:
The emergence of multidimensional data presents significant challenges for traditional regression models based on matrices or vectors, particularly in capturing multi-directional correlations. In response, tensor regression has been proposed as a powerful framework for modeling linear relationships among multi-dimensional variables. In this paper, we introduce a high-dimensional tensor-response tensor regression model under low-dimensional structural assumptions, such as sparsity and low-rankness. Assuming the underlying tensor lies within an unknown low-dimensional subspace, we consider a least squares estimation framework with non-convex penalties. Theoretically, we derive general risk bounds for the resulting estimators and demonstrate that they achieve the oracle statistical rates under mild technical conditions. To compute the proposed estimators efficiently, we introduce an accelerated proximal gradient algorithm demonstrating rapid convergence in practice. Extensive experiments on synthetic and real-world datasets validate the effectiveness of the proposed regression model and showcase the practical utility of the theoretical findings.

Paperid:377

Authors:Tianjian Li · Daniel Khashabi

Title: SIMPLEMIX: Frustratingly Simple Mixing of Off- and On-policy Data in Language Model Preference Learning

Abstract:
Aligning language models with human preferences relies on pairwise preference datasets. While some studies suggest that onpolicy data consistently outperforms off-policy data for preference learning, others indicate that the advantages of on-policy data are task-dependent, highlighting the need for a systematic exploration of their interplay.In this work, we show that on-policy and off-policy data offer complementary strengths: on-policy data is particularly effective for reasoning tasks like math and coding, while off-policy data performs better on subjective tasks such as creative writing and making personal recommendations. Guided by these findings, we introduce SimpleMix, an approach to combine the complementary strengths of on-policy and off-policy preference learning by simply mixing these two data sources. Our empirical results across diverse tasks and benchmarks demonstrate that SimpleMix substantially improves language model alignment. Specifically, SimpleMix improves upon on-policy DPO and off-policy DPO by an average of 6.03 on Alpaca Eval 2.0. Moreover, it surpasses prior approaches that are much more complex in combining on- and off-policy data, such as HyPO and DPO-Mix-P, by an average of 3.05. These findings validate the effectiveness and efficiency of SimpleMix for enhancing preference-based alignment.

Paperid:378

Authors:Dong Li · Yidi Liu · Xueyang Fu · Jie Huang · Senyan Xu · Qi Zhu · Zheng-Jun Zha

Title: FourierMamba: Fourier Learning Integration with State Space Models for Image Deraining

Abstract:
Image deraining aims to remove rain streaks from rainy images and restore clear backgrounds. Currently, some research that employs the Fourier transform has proved to be effective for image deraining, due to it acting as an effective frequency prior for capturing rain streaks. However, despite there exists dependency of low frequency and high frequency in images, these Fourierbased methods rarely exploit the correlation of different frequencies for conjuncting their learning procedures, limiting the full utilization of frequency information for image deraining. Alternatively, the recently emerged Mamba technique depicts its effectiveness and efficiency for modeling correlation in various domains (e.g., spatial, temporal), and we argue that introducing Mamba into its unexplored Fourier spaces to correlate different frequencies would help improve image deraining. This motivates us to propose a new framework termed FourierMamba, which performs image deraining with Mamba in the Fourier space. Owing to the unique arrangement of frequency orders in Fourier space, the core of FourierMamba lies in the scanning encoding of different frequencies, where the low-high frequency order formats exhibit differently in the spatial dimension (unarranged in axis) and channel dimension (arranged in axis). Therefore, we design FourierMamba that correlates Fourier space information in the spatial and channel dimensions with distinct designs. Specifically, in the spatial dimension Fourier space, we introduce the zigzag coding to scan the frequencies to rearrange the orders from low to high frequencies, thereby orderly correlating the connections between frequencies; in the channel dimension Fourier space with arranged orders of frequencies in axis, we can directly use Mamba to perform frequency correlation and improve the channel information representation. Extensive experiments reveal that our method outperforms state-of-the-art methods both qualitatively and quantitatively.

Paperid:379

Authors:Rohan Shenoy · Evan Coleman · Hans Gaensbauer · Elsa Olivetti

Title: Counting atoms faster: policy-based nuclear magnetic resonance pulse sequencing for atomic abundance measurement

Abstract:
Quantifying the elemental composition of a material is a general scientific challenge with broad relevance to environmental sustainability. Existing techniques for the measurement of atomic abundances generally require laboratory conditions and expensive equipment. As a result, they cannot be deployedin situwithout significant capital investment, limiting their proliferation. Measurement techniques based on nuclear magnetic resonance (NMR) hold promise in this setting due to their applicability across the periodic table, their nondestructive manipulation of samples, and their amenability toin silicooptimization. In this work, we learn policies to modulate NMR pulses for rapid atomic abundance quantification. Our approach involves three inter-operating agents which (1) rapidly align nuclear spins for measurement, (2) quickly force relaxation to equilibrium, and (3) toggle control between agents (1) and (2) to minimize overall measurement time. To demonstrate this technique, we consider a specific use case of low-magnetic-field carbon-13 quantification for low-cost, portable analysis of foodstuffs and soils. We find significant performance improvements relative to traditional NMR pulse sequencing, and discuss limitations on the applicability of this approach.

Paperid:380

Authors:Jin Zhu · Jingyi Li · Hongyi Zhou · Yinan Lin · Chengchun Shi · Zhenhua Lin

Title: Balancing Interference and Correlation in Spatial Experimental Designs: A Causal Graph Cut Approach

Abstract:
This paper focuses on the design of spatial experiments to optimize the amount of information derived from the experimental data and enhance the accuracy of the resulting causal effect estimator. We propose a surrogate function for the mean squared error (MSE) of the estimator, which facilitates the use of classical graph cut algorithms to learn the optimal design. Our proposal offers three key advances: (1) it accommodates moderate to large spatial interference effects; (2) it adapts to different spatial covariance functions; (3) it is computationally efficient. Theoretical results and numerical experiments based on synthetic environments and a dispatch simulator that models a cityscale ridesharing market, further validate the effectiveness of our design.

Paperid:381

Authors:Yuxuan Sun · Ruikang Liu · Haoli Bai · Han Bao · Kang Zhao · Yuening Li · JiaxinHu · Xianzhi Yu · Lu Hou · Chun Yuan · Xin Jiang · Wulong Liu · Jun Yao

Title: FlatQuant: Flatness Matters for LLM Quantization

Abstract:
Recently, quantization has been widely used for the compression and acceleration of large language models (LLMs). Due to the outliers in LLMs, it is crucial to flatten weights and activations to minimize quantization error with equally spaced quantization points. Prior research explores various prequantization transformations to suppress outliers, such as per-channel scaling and Hadamard transformation. However, we observe that these transformed weights and activations can still exhibit steep and dispersed distributions. In this paper, we propose FlatQuant (Fast and Learnable Affine Transformation), a new post-training quantization approach that enhances the flatness of weights and activations. Our approach identifies optimal affine transformations tailored to each linear layer, calibrated in hours via a lightweight objective. To reduce runtime overhead, we apply Kronecker decomposition to the transformation matrices, and fuse all operations in FlatQuant into a single kernel. Extensive experiments demonstrate that FlatQuant establishes a new state-of-the-art benchmark for quantization. For example, the proposed method achieves less than 1\% accuracy drop for W4A4 quantization on the LLaMA-3-70B model, surpassing SpinQuant by 7.5\%. Additionally, it provides up to 2.3x prefill speedup and 1.7x decoding speedup compared to the FP16 model.

Paperid:382

Authors:Bilgehan Sel · Lifu Huang · Naren Ramakrishnan · Ruoxi Jia · Ming Jin

Title: LLMs Can Plan Faster Only If We Let Them

Abstract:
Large language models (LLMs) are making inroads into classical AI problems such as automated planning, yet key shortcomings continue to hamper their integration. Chainof-Thought (CoT) struggles in complex multi-step reasoning, and Tree-of-Thoughts requires multiple queries that increase computational overhead. Recently, Algorithm-of-Thoughts (AoT) have shown promise using in-context examples, at the cost of significantly longer solutions compared to CoT. Aimed at bridging the solution length gap between CoT and AoT, this paper introduces AoT-O3, which combines supervised finetuning on AoT-style plans with a reinforcement learning (RL) framework designed to reduce solution length. The RL component uses a reward model that favors concise, valid solutions while maintaining planning accuracy. Empirical evaluations indicate that AoT-O3 shortens solution length by up to 80\% compared to baseline AoT while maintaining or surpassing prior performance. These findings suggest a promising pathway for more efficient, scalable LLM-based planning.

Paperid:383

Authors:Hao Li · Yu-Hao Huang · Chang Xu · Viktor Schlegel · Renhe Jiang · Riza Batista-Navarro · Goran Nenadic · Jiang Bian

Title: BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimisation and Diffusion Modelling

Abstract:
Timeseries Generation (TSG) is a prominent research area with broad applications in simulations, data augmentation, and counterfactual analysis. While existing methods have shown promise in unconditional single-domain TSG, real-world applications demand for cross-domain approaches capable of controlled generation tailored to domain-specific constraints and instance-level requirements. In this paper, we argue that text can provide semantic insights, domain information and instance-specific temporal patterns, to guide and improve TSG. We introduce ``Text-Controlled TSG'', a task focused on generating realistic time series by incorporating textual descriptions.To address data scarcity in this setting, we propose a novel LLM-based Multi-Agent framework that synthesizes diverse, realistic text-to-TS datasets. Furthermore, we introduce Bridge, a hybrid text-controlled TSG framework that integrates semantic prototypes with text description for supporting domain-level guidance. This approach achieves state-of-the-art generation fidelity on 11 of 12 datasets, and improves controllability by 12.52% on MSE and 6.34% MAE compared to no text input generation, highlighting its potential for generating tailored time-series data

Paperid:384

Authors:Georgy Noarov · Ramya Ramalingam · Aaron Roth · Stephan Xie

Title: High-Dimensional Prediction for Sequential Decision Making

Abstract:
Abstract:We give an efficient algorithm for producing multidimensional forecasts in an online adversarial environment that have low bias subject to any polynomial number of conditioning events, that can depend both on external context and on our predictions themselves. We demonstrate the use of this algorithm with several applications. We show how to make predictions that can be transparently consumed by any polynomial number of downstream decision makers with different utility functions, guaranteeing them diminishing swap regret at optimal rates. We also give the first efficient algorithms for guaranteeing diminishing conditional regret in online combinatorial optimization problems for an arbitrary polynomial number of conditioning events --- i.e. on an arbitrary number of intersecting subsequences determined both by context and our own predictions. Finally we give the first efficient algorithm guaranteeing online multicalibration with $T^{2/3}$ rates in the ECE metric.

Paperid:385

Authors:Michael Sander · Vincent Roulet · Tianlin Liu · Mathieu Blondel

Title: Joint Learning of Energy-based Models and their Partition Function

Abstract:
Energybased models (EBMs) offer a flexible framework for parameterizing probability distributions using neural networks.However, learning EBMs by exact maximum likelihood estimation (MLE) is generally intractable, due to the need to compute the partition function.In this paper, we propose a novel min-min formulation for approximately learning probabilistic EBMs in combinatorially-large discrete spaces, such as sets or permutations. Our key idea is to jointly learn both an energy model and its log-partition, parameterized as a neural network. Our approach not only provides a novel tractable objective criterion to learn EBMs by stochastic gradient descent (without relying on MCMC), but also a novel means to estimate the log-partition function on unseen data points.On the theoretical side, we show that our approach recovers the optimal MLE solution when optimizing in the space of continuous functions.Furthermore, we show that our approach naturally extends to the broader family of Fenchel-Young losses, allowing us to obtainthe first tractable method for optimizing the sparsemax loss in combinatorially-large spaces.We demonstrate our approach on multilabel classification and label ranking.

Paperid:386

Authors:Zhigaoyuan Wang · Ying Sun · Hengshu Zhu

Title: Unifying Knowledge from Diverse Datasets to Enhance Spatial-Temporal Modeling: A Granularity-Adaptive Geographical Embedding Approach

Abstract:
Spatiotemporal forecasting provides potential for discovering evolutionary patterns in geographical scientific data. However, geographical scientific datasets are often manually collected across studies, resulting in limited time spans and data scales. This hinders existing methods that rely on rich historical data for individual entities. In this paper, we argue that heterogeneous datasets from different studies can provide complementary insights into the same underlying system, helping improve predictions for geographical entities with limited historical data. To this end, we propose a Segment Quadtree Geographical Embedding Framework (SQGEF). SQGEF integrates knowledge from datasets with varied target entities, time spans, and observation variables to learn unified representations for multi-granularity entities—including those absent during training. Specifically, we propose a novel data structure, Segment Quadtree, that flexibly accommodates entities of varying granularities. SQGEF not only captures multi-level interactions from grid data but also extracts nested relationships and human-defined boundaries from diverse entities, enabling a comprehensive understanding of complex geographical structures. Experiments on real-world datasets demonstrate that SQGEF effectively represents unseen geographical entities and enhances performance for various models.

Paperid:387

Authors:Zhengqi Gao · Kaiwen Zha · Tianyuan Zhang · Zihui Xue · Duane Boning

Title: REG: Rectified Gradient Guidance for Conditional Diffusion Models

Abstract:
Guidance techniques are simple yet effective for improving conditional generation in diffusion models. Albeit their empirical success, the practical implementation of guidance diverges significantly from its theoretical motivation. In this paper, we reconcile this discrepancy by replacing the scaled marginal distribution target, which we prove theoretically invalid, with a valid scaled joint distribution objective. Additionally, we show that the established guidance implementations are approximations to the intractable optimal solution under no future foresight constraint. Building on these theoretical insights, we propose rectified gradient guidance (REG), a versatile enhancement designed to boost the performance of existing guidance methods. Experiments on 1D and 2D demonstrate that REG provides a better approximation to the optimal solution than prior guidance techniques, validating the proposed theoretical framework. Extensive experiments on classconditional ImageNet and text-to-image generation tasks show that incorporating REG consistently improves FID and Inception/CLIP scores across various settings compared to its absence.

Paperid:388

Authors:Ruizhe Wang · Yeyun Gong · Xiao Liu · Guoshuai Zhao · Ziyue Yang · Baining Guo · Zheng-Jun Zha · Peng CHENG

Title: Optimizing Large Language Model Training Using FP4 Quantization

Abstract:
The growing computational demands of training large language models (LLMs) necessitate more efficient methods. Quantized training presents a promising solution by enabling lowbit arithmetic operations to reduce these costs. While FP8 precision has demonstrated feasibility, leveraging FP4 remains a challenge due to significant quantization errors and limited representational capacity. This work introduces the first FP4 training framework for LLMs, addressing these challenges with two key innovations: a differentiable quantization estimator for precise weight updates and an outlier clamping and compensation strategy to prevent activation collapse. To ensure stability, the framework integrates a mixed-precision training scheme and vector-wise quantization. Experimental results demonstrate that our FP4 framework achieves accuracy comparable to BF16 and FP8, with minimal degradation, scaling effectively to 13B-parameter LLMs trained on up to 100B tokens. With the emergence of next-generation hardware supporting FP4, our framework sets a foundation for efficient ultra-low precision training.

Paperid:389

Authors:Jueqing Lu · Wray Buntine · Yuanyuan Qi · Joanna Dipnall · Belinda Gabbe · Lan Du

Title: Navigating Conflicting Views: Harnessing Trust For Learning

Abstract:
Resolving conflicts is essential to make the decisions of multiview classification more reliable. Much research has been conducted on learning consistent and informative representations among different views, often assuming that all views are equally important and perfectly aligned. However, real-world multi-view data may not always conform to these assumptions, as some views may express distinct information. To address this issue, we develop a computational trust-based discounting method to enhance the existing Evidential Multi-view framework in scenarios where conflicts between different views may arise. Its belief fusion process considers the reliability of predictions made by individual views via an instance-wise probability-sensitive trust discounting mechanism.We evaluate our method on six real-world datasets, using Top-1 Accuracy, Fleiss’ Kappa, and a new metric called Multi-View Agreement with Ground Truth that takes into consideration the ground truth labels, to measure the reliability of the prediction.We also evaluate whether uncertainty measures can effectively indicate prediction correctness by calculating the AUROC.The experimental results show that computational trust can effectively resolve conflicts, paving the way for more reliable multi-view classification models in real-world applications.

Paperid:390

Authors:Feng Wang · Yaodong Yu · Wei Shao · Yuyin Zhou · Alan Yuille · Cihang Xie

Title: Scaling Laws in Patchification: An Image Is Worth 50,176 Tokens And More

Abstract:
Since the introduction of Vision Transformer (ViT), patchification has long been regarded as a common image preprocessing approach for plain visual architectures. By compressing the spatial size of images, this approach can effectively shorten the token sequence and reduce the computational cost of ViT-like plain architectures. In this work, we aim to thoroughly examine the information loss caused by this patchification-based compressive encoding paradigm and how it affects visual understanding. We conduct extensive patch size scaling experiments and excitedly observe an intriguing scaling law in patchification: the models can consistently benefit from decreased patch sizes and attain improved predictive performance, until it reaches the minimum patch size of 1*1, i.e., pixel tokenization. This conclusion is broadly applicable across different vision tasks, various input scales, and diverse architectures such as ViT and the recent Mamba models. Moreover, as a by-product, we discover that with smaller patches, task-specific decoder heads become less critical for dense prediction. In the experiments, we successfully scale up the visual sequence to an exceptional length of 50,176 tokens, achieving a competitive test accuracy of 84.6% with a base-sized model on the ImageNet-1k benchmark. We hope this study can provide insights and theoretical foundations for future works of building non-compressive vision models.

Paperid:391

Authors:Martin Van Waerebeke · Kevin Scaman · Marco Lorenzi · Giovanni Neglia

Title: When to Forget? Complexity Trade-offs in Machine Unlearning

Abstract:
Machine Unlearning (MU) aims at removing the influence of specific data points from a trained model, striving to achieve this at a fraction of the cost of full model retraining. In this paper, we analyze the efficiency of unlearning methods and establish the first upper and lower bounds on minimax computation times for this problem, characterizing the performance of the most efficient algorithm against the most difficult objective function. Specifically, for strongly convex objective functions and under the assumption that the forget data is inaccessible to the unlearning method, we provide a phase diagram for the unlearning complexity ratioa novel metric that compares the computational cost of the best unlearning method to full model retraining. The phase diagram reveals three distinct regimes: one where unlearning at a reduced cost is infeasible, another where unlearning is trivial because adding noise suffices, and a third where unlearning achieves significant computational advantages over retraining. These findings highlight the critical role of factors such as data dimensionality, the number of samples to forget, and privacy constraints in determining the practical feasibility of unlearning.

Paperid:392

Authors:Jiangshan Wang · Junfu Pu · Zhongang Qi · Jiayi Guo · Yue Ma · Nisha Huang · Yuxin Chen · Xiu Li · Ying Shan

Title: Taming Rectified Flow for Inversion and Editing

Abstract:
Rectifiedflow-based diffusion transformers like FLUX and OpenSora have demonstrated outstanding performance in the field of image and video generation. Despite their robust generative capabilities, these models often struggle with inversion inaccuracies, which could further limit their effectiveness in downstream tasks such as image and video editing. To address this issue, we propose RF-Solver, a novel training-free sampler that effectively enhances inversion precision by mitigating the errors in the ODE-solving process of rectified flow. Specifically, we derive the exact formulation of the rectified flow ODE and apply the high-order Taylor expansion to estimate its nonlinear components, significantly enhancing the precision of ODE solutions at each timestep. Building upon RF-Solver, we further propose RF-Edit, a general feature-sharing-based framework for image and video editing. By incorporating self-attention features from the inversion process into the editing process, RF-Edit effectively preserves the structural information of the source image or video while achieving high-quality editing results. Our approach is compatible with any pre-trained rectified-flow-based models for image and video tasks, requiring no additional training or optimization. Extensive experiments across generation, inversion, and editing tasks in both image and video modalities demonstrate the superiority and versatility of our method. The source code will be released.

Paperid:393

Authors:Minting Pan · Yitao Zheng · Jiajian Li · Yunbo Wang · Xiaokang Yang

Title: Video-Enhanced Offline Reinforcement Learning: A Model-Based Approach

Abstract:
Abstract:Offline reinforcement learning (RL) optimizes control policies using static datasets, reducing the need for costly and potentially unsafe exploration in realworld scenarios. However, it faces key challenges, particularly value overestimation due to the lack of an interactive environment. In this paper, we introduce Video-Enhanced Offline RL (VeoRL), a novel model-based RL approach that builds an interactive world model using a diverse collection of unlabeled, heterogeneous video data—freely accessible and potentially unlimited from the Internet. Through model-based policy optimization, VeoRL effectively transfers commonsense knowledge of control behaviors and physical dynamics, extracted from natural videos, to the RL agent. Our approach demonstrates substantial improvements over existing offline RL methods across a wide range of visuomotor control tasks, including robotic manipulation, autonomous driving, and open-world video games, delivering over a $100\\%$ performance boost on multiple tasks.

Paperid:394

Authors:Yifei Zhou · Qianlan Yang · Kaixiang Lin · Min Bai · Xiong Zhou · Yu-Xiong Wang · Sergey Levine · Li Li

Title: Proposer-Agent-Evaluator (PAE): Autonomous Skill Discovery For Foundation Model Internet Agents

Abstract:
A generalist foundation model agent needs to have a large and diverse skill repertoire, such as finding directions between two travel locations and buying specific items from the Internet. If each skill needs to be specified manually through a fixed set of humanannotated instructions, the agent’s skill repertoire will necessarily be limited due to the scalability of human-annotated instructions. In this work, we address this challenge by proposing Proposer-Agent-Evaluator (PAE), an effective learning system that enables foundation model agents to autonomously discover and practice skills in the wild. After a context-aware task proposer generates instructions based on website information, the agent policy attempts those tasks in the real world with resulting trajectories evaluated by an autonomous VLM-based success evaluator. The success evaluation serves as the reward signal for the agent to refine its policies through RL. We validate PAE on challenging vision-based web navigation, using both real-world and selfhosted websites from WebVoyager and WebArena. Our results show that PAE significantly improves the zero-shot generalization capability of VLM Internet agents (around 50% relative improvement)to both unseen tasks and websites.

Paperid:395

Authors:Severi Rissanen · RuiKang OuYang · Jiajun He · Wenlin Chen · Markus Heinonen · Arno Solin · Jose Miguel Hernandez-Lobato

Title: Progressive Tempering Sampling with Diffusion

Abstract:
Recent research has focused on designing neural samplers that amortize the process of sampling from unnormalized densities. However, despite significant advancements, they still fall short of the stateof-the-art MCMC approach, Parallel Tempering (PT), in terms of the number of target evaluations. On the other hand, we also identify several inefficiencies in PT: (1) information across different temperature levels is only exchanged through swapping samples; (2) the varying mixing rates across temperatures lead to wasted computation, particularly at low temperatures early on; and (3) PT’s reliance on MCMC results in dependent samples, slowing transitions between modes. To address the weaknesses in both approaches, we propose Progressive Tempering Sampler with Diffusion (PTSD), which trains diffusion models sequentially across temperatures. We introduce a novel method to combine high-temperature diffusion models to generate approximate lower-temperature samples, which are minimally refined using MCMC and used to train the next diffusion model. This approach enables efficient reuse of sample information across temperature levels while generating well-mixed, uncorrelated samples. Our method significantly improves target evaluation efficiency,outperforming diffusion-based neural samplers and demonstrating a promising direction to improve over PT.

Paperid:396

Authors:Boyuan Li · Yicheng Luo · Zhen Liu · Junhao Zheng · Jianming Lv · Qianli Ma

Title: HyperIMTS: Hypergraph Neural Network for Irregular Multivariate Time Series Forecasting

Abstract:
Irregular multivariate time series (IMTS) are characterized by irregular time intervals within variables and unaligned observations across variables, posing challenges in learning temporal and variable dependencies. Many existing IMTS models either require padded samples to learn separately from temporal and variable dimensions, or represent original samples via bipartite graphs or sets.However, the former approaches often need to handle extra padding values affecting efficiency and disrupting original sampling patterns, while the latter ones have limitations in capturing dependencies among unaligned observations.To represent and learn both dependencies from original observations in a unified form, we propose HyperIMTS, aHypergraph neural network forIrregularMultivariateTimeSeries forecasting.Observed values are converted as nodes in the hypergraph, interconnected by temporal and variable hyperedges to enable message passing among all observations.Through irregularityaware message passing, HyperIMTS captures variable dependencies in a time-adaptive way to achieve accurate forecasting. Experiments demonstrate HyperIMTS's competitive performance among state-of-the-art models in IMTS forecasting with low computational cost.

Paperid:397

Authors:Hyeonah Kim · Sanghyeok Choi · Jiwoo Son · Jinkyoo Park · Changhyun Kwon

Title: Neural Genetic Search in Discrete Spaces

Abstract:
Effective search methods are crucial for improving the performance of deep generative models at test time. In this paper, we introduce a novel testtime search method, Neural Genetic Search (NGS), which incorporates the evolutionary mechanism of genetic algorithms into the generation procedure of deep models. The core idea behind NGS is its crossover, which is defined as parent-conditioned generation using trained generative models. This approach offers a versatile and easy-to-implement search algorithm for deep generative models. We demonstrate the effectiveness and flexibility of NGS through experiments across three distinct domains: routing problems, adversarial prompt generation for language models, and molecular design.

Paperid:398

Authors:Haocheng Xi · Shuo Yang · Yilong Zhao · Chenfeng Xu · Muyang Li · Xiuyu Li · Yujun Lin · Han Cai · Jintao Zhang · Dacheng Li · Jianfei Chen · Ion Stoica · Kurt Keutzer · Song Han

Title: Sparse Video-Gen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity

Abstract:
Abstract:Diffusion Transformers (DiTs) dominate video generation but their high computational cost severely limits realworld applicability, usually requiring tens of minutes to generate a few seconds of video even on high-performance GPUs. This inefficiency primarily arises from the quadratic computational complexity of 3D full attention with respect to the context length. In this paper, we propose a training-free framework termed Sparse VideoGen (SVG) that leverages the inherent sparsity in 3D full attention to boost inference efficiency. We reveal that the attention heads can be dynamically classified into two groups depending on distinct sparse patterns: (1) Spatial Head, where only spatially-related tokens within each frame dominate the attention output, and (2) Temporal Head, where only temporally-related tokens across different frames dominate. Based on this insight, SVG proposes an online profiling strategy to capture the dynamic sparse patterns and predicts the type of attention head. Combined with a novel hardware-efficient tensor layout transformation and customized kernel implementations, SVG achieves up to 2.28$\times$ and 2.33$\times$ end-to-end speedup on CogVideoX-v1.5 and HunyuanVideo, respectively, while preserving generation quality. Our code will be open-sourced upon publication.

Paperid:399

Authors:Haoyun Jiang · Haolin li · jianwei zhang · Fei Huang · Qiang Hu · Minmin Sun · Shuai Xiao · Yong Li · Junyang Lin · Jiangchao Yao

Title: CateKV: On Sequential Consistency for Long-Context LLM Inference Acceleration

Abstract:
Abstract:Large language models (LLMs) have demonstrated strong capabilities in handling longcontext tasks, but processing such long contexts remains challenging due to the substantial memory requirements and inference latency. In this work, we discover that certain attention heads exhibit sequential consistency in their attention patterns, which can be persistently identified using a coefficient-of-variation-based algorithm. Inspired by this observation, we propose CateKV, a hybrid KV cache method that retains only critical token information for consistent heads, thereby reducing KV cache size and computational overhead, while preserving the majority of KV pairs in adaptive heads to ensure high accuracy. We show the unique characteristics of our algorithm and its extension with existing acceleration methods. Comprehensive evaluations on long-context benchmarks show that, while maintaining accuracy comparable to full attention, CateKV reduces memory usage by up to $2.72\times$ and accelerates decoding by $2.18\times$ in single-sample inputs, and boosts throughput by $3.96\times$ in batch scenarios.

Paperid:400

Authors:Shentong Mo · Sukmin Yun

Title: GMAIL: Generative Modality Alignment for generated Image Learning

Abstract:
Generative models have made it possible to synthesize highly realistic images, potentially providing an abundant data source for training machine learning models. Despite the advantages of these synthesizable data sources, the indiscriminate use of generated images as real images for training can even cause mode collapse due to modality discrepancies between real and synthetic domains. In this paper, we propose a novel framework for discriminative use of generated images, coined \textit{GMAIL}, that explicitly treats generated images as a separate modality from real images. Instead of indiscriminately replacing real images with generated ones in the pixel space, our approach bridges the two distinct modalities in the same latent space through a multimodal learning approach. To be specific, we first fine-tune a model exclusively on generated images using a cross-modality alignment loss and then employ this aligned model to further train various vision-language models with generated images. By aligning the two modalities, our approach effectively leverages the benefits of recent advances in generative models, thereby boosting the effectiveness of generated image learning across a range of vision-language tasks. Our framework can be easily incorporated with various vision-language models, and we demonstrate its efficacy throughout extensive experiments. For example, our framework significantly improves performance on image captioning, zero-shot image retrieval, zero-shot image classification, and long caption retrieval tasks. It also shows positive generated data scaling trends and notable enhancements in the captioning performance of the large multimodal model, LLaVA.

Paperid:401

Authors:Jinghang Yue · Jing Wang · Lu Zhang · Shuo Zhang · Da Li · Zhaoyang Ma · Youfang Lin

Title: Optimal Information Retention for Time-Series Explanations

Abstract:
Explaining deep models for timeseries data is crucial for identifying key patterns in sensitive domains, such as healthcare and finance. However, due to the lack of unified optimization criterion, existing explanation methods often suffer from redundancy and incompleteness, where irrelevant patterns are included or key patterns are missed in explanations. To address this challenge, we propose the Optimal Information Retention Principle, where conditional mutual information defines minimizing redundancy and maximizing completeness as optimization objectives. We then derive the corresponding objective function theoretically. As a practical framework, we introduce an explanation framework ORTE, learning a binary mask to eliminate redundant information while mining temporal patterns of explanations. We decouple the discrete mapping process to ensure the stability of gradient propagation, while employing contrastive learning to achieve precise filtering of explanatory patterns through the mask, thereby realizing a trade-off between low redundancy and high completeness. Extensive quantitative and qualitative experiments on synthetic and real-world datasets demonstrate that the proposed principle significantly improves the accuracy and completeness of explanations compared to baseline methods. The code is available at*.

Paperid:402

Authors:Meng Chen · Hongwei Jia · Zechen Li · Weiming Huang · Wenzhen Jia · Kai Zhao · Hongjun Dai

Title: Cross-City Latent Space Alignment for Consistency Region Embedding

Abstract:
Urban region embeddings have recently seen significant progress. However, existing methods remain largely restricted to one city, which limits crosscity knowledge transfer and the re-usage of them for downstream tasks (e.g., population estimation) in other cities. To overcome this limitation, we propose the Consistency Region Embedding (CoRE), a unified framework that integrates region embedding learning with cross-city latent space alignment. CoRE embeds regions from different cities into distinct latent spaces, then aligns their latent space manifolds while ensuring fine-grained compatibility. This alignment enables region representations from different cities to be directly applied within a shared latent space, facilitating the re-usage of the model for other cities. Extensive experiments demonstrate that CoRE outperforms the state-of-the-arts baselines, underscoring its effectiveness in enabling knowledge transfer through aligned latent spaces.

Paperid:403

Authors:Fan Xu · Cheng Xin · Xin Ding · Jie Gao · Jiaxin Ding

Title: TopInG: Topologically Interpretable Graph Learning via Persistent Rationale Filtration

Abstract:
Graph Neural Networks (GNNs) have shown remarkable performance in various scientific domains, but their lack of interpretability limits their applicability in critical decisionmaking processes. Recently, intrinsic interpretable GNNs have been studied to provide insights into model predictions by identifying rationale substructures in graphs. However, existing methods face challenges when the underlying rationale subgraphs are complicated and variable. To address this challenge, we propose TOPING, a novel topological framework to interpretable GNNs that leverages persistent homology to identify persistent rationale subgraphs. Our method introduces a rationale filtration learning technique that models the generating procedure of rationale subgraphs, and enforces the persistence of topological gap between rationale subgraphs and complement random graphs by a novel self-adjusted topological constraint, topological discrepancy. We show that our topological discrepancy is a lower bound of a Wasserstein distance on graph distributions with Gromov-Hausdorff metric. We provide theoretical guarantees showing that our loss is uniquely optimized by the ground truth under certain conditions. Through extensive experiments on various synthetic and real datasets, we demonstrate that TOPING effectively addresses key challenges in interpretable GNNs including handling variiform rationale subgraphs, balancing performance with interpretability, and avoiding spurious correlations. Experimental results show that our approach improves state-of-the-art on both predictive accuracy and interpretation quality.

Paperid:404

Authors:Harikrishna Metta · Venkatesh Babu Radhakrishnan

Title: Discovering a Zero (Zero-Vector Class of Machine Learning)

Abstract:
In Machine learning, separating data into classes is a very fundamental problem. A mathematical framework around the classes is presented in this work to deepen the understanding of classes. The classes are defined as vectors in a Vector Space, where Addition corresponds to the union of classes, and scalar multiplication resembles set complement of classes. The ZeroVector in the vector space corresponds to a class called Zero-Vector Class. This discovery enables numerous applications. One such application, termed 'clear learning' in this work, focuses on learning the true nature (manifold) of the data instead of merely learning a boundary sufficient for classification. Another application, called 'unary class learning', involves learning a single class in isolation rather than learning by comparing two or more classes. Additionally, 'set operations on classes' is another application highlighted in this work. Furthermore, Continual Learning of classes is facilitated by smaller networks. The Zero-Vector Class enables neural networks to learn only the data manifold; therefore, it can also used for generation of new data. Results for the key applications are shown using the MNIST dataset. To further strengthen the claims, some results are also produced using the CIFAR-10 dataset.

Paperid:405

Authors:Ahmad Beirami · Alekh Agarwal · Jonathan Berant · Alexander D'Amour · Jacob Eisenstein · Chirag Nagpal · Ananda Suresh

Title: Theoretical guarantees on the best-of-n alignment policy

Abstract:
Abstract:A simple and effective method for the inferencetime alignment of generative models is the best-of-$n$ policy, where $n$ samples are drawn from a reference policy, ranked based on a reward function, and the highest ranking one is selected. A commonly used analytical expression in the literature claims that the KL divergence between the best-of-$n$ policy and the reference policy is equal to $\log (n) - (n-1)/n.$ We disprove the validity of this claim, and show that it is an upper bound on the actual KL divergence. We also explore the tightness of this upper bound in different regimes, and propose a new estimator for the KL divergence and empirically show that it provides a tight approximation. We also show that the win rate of the best-of-$n$ policy against the reference policy is upper bounded by $n/(n+1)$ and derive bounds on the tightness of this characterization. We conclude with analyzing the tradeoffs between win rate and KL divergence of the best-of-$n$ alignment policy, which demonstrate that very good tradeoffs are achievable with $n < 1000$.

Paperid:406

Authors:Mao-Lin Luo · Zi-Hao Zhou · Tong Wei · Min-Ling Zhang

Title: LADA: Scalable Label-Specific CLIP Adapter for Continual Learning

Abstract:
Continual learning with visionlanguage models like CLIP offers a pathway toward scalable machine learning systems by leveraging its transferable representations. Existing CLIP-based methods adapt the pre-trained image encoder by adding multiple sets of learnable parameters, with each task using a partial set of parameters. This requires selecting the expected parameters for input images during inference, which is prone to error that degrades performance. To address this problem, we introduce LADA (Label-specificADApter). Instead of partitioning parameters across tasks, LADA appends lightweight, label-specific memory units to the frozen CLIP image encoder, enabling discriminative feature generation by aggregating task-agnostic knowledge. To prevent catastrophic forgetting, LADA employs feature distillation for seen classes, preventing their features from interference by new classes. Positioned after the image encoder, LADA prevents gradient flow to the frozen CLIP parameters, ensuring efficient training. Extensive results show that LADA achieves state-of-the-art performance in continual learning settings. The supplementary material includes source code for reproducibility.

Paperid:407

Authors:Jiaqi Zhu · Shaofeng Cai · Shen · Gang Chen · Fang Deng · Beng Chin Ooi

Title: In-Context Adaptation to Concept Drift for Learned Database Operations

Abstract:
Abstract:Machine learning has demonstrated transformative potential for database operations, such as query optimization and indatabase data analytics. However, dynamic database environments, characterized by frequent updates and evolving data distributions, introduce concept drift, which leads to performance degradation for learned models and limits their practical applicability. Addressing this challenge requires efficient frameworks capable of adapting to shifting concepts while minimizing the overhead of retraining or fine-tuning.In this paper, we propose FLAIR, an online adaptation framework that introduces a new paradigm called \textit{in-context adaptation} for learned database operations. FLAIR leverages the inherent property of data systems, i.e., immediate availability of execution results for predictions, to enable dynamic context construction. By formalizing adaptation as $f:(\mathbf{x} | \mathcal{C}_t) \to \mathbf{y}$, with $\mathcal{C}_t$ representing a dynamic context memory, FLAIR delivers predictions aligned with the current concept, eliminating the need for runtime parameter optimization. To achieve this, FLAIR integrates two key modules: a Task Featurization Module for encoding task-specific features into standardized representations, and a Dynamic Decision Engine, pre-trained via Bayesian meta-training, to adapt seamlessly using contextual information at runtime. Extensive experiments across key database tasks demonstrate that FLAIR outperforms state-of-the-art baselines, achieving up to $5.2\times$ faster adaptation and reducing error by 22.5\% for cardinality estimation.

Paperid:408

Authors:Christian Lange · Max Hamilton · Elijah Cole · Alexander Shepard · Samuel Heinrich · Angela Zhu · Subhransu Maji · Grant Horn · Oisin Mac Aodha

Title: Few-shot Species Range Estimation

Abstract:
Knowing where a particular species can or cannot be found on Earth is crucial for ecological research and conservation efforts. By mapping the spatial ranges of all species, we would obtain deeper insights into how global biodiversity is affected by climate change and habitat loss. However, accurate range estimates are only available for a relatively small proportion of all known species. For the majority of the remaining species, we often only have a small number of records denoting the spatial locations where they have previously been observed. We outline a new approach for fewshot species range estimation to address the challenge of accurately estimating the range of a species from limited data. During inference, our model takes a set of spatial locations as input, along with optional metadata such as text or an image, and outputs a species encoding that can be used to predict the range of a previously unseen species in feed-forward manner. We validate our method on two challenging benchmarks, where we obtain state-of-the-art range estimation performance, in a fraction of the compute time, compared to recent alternative approaches.

Paperid:409

Authors:Daniel Zilberg · Ron Levie

Title: PieClam: A Universal Graph Autoencoder Based on Overlapping Inclusive and Exclusive Communities

Abstract:
We propose PieClam (Prior Inclusive Exclusive Cluster Affiliation Model): a graph autoencoder, where nodes are embedded into a code space by an algorithm that maximizes the loglikelihood of the decoded graph. PieClam is a community affiliation model that extends well-known methods like BigClam in two main manners. First, instead of the decoder being defined via pairwise interactions between the nodes in the code space, we also incorporate a learned prior on the distribution of nodes in the code space, turning our method into a graph generative model. Secondly, we generalize the notion of communities by allowing not only sets of nodes with strong connectivity, which we call inclusive communities, but also sets of nodes with strong disconnection, which we call exclusive communities. By introducing a new graph similarity measure, called the log cut distance, we show that PieClam is a universal autoencoder, able to uniformly approximately reconstruct any graph. Our method is shown to obtain competitive performance in graph anomaly detection and link prediction benchmarks.

Paperid:410

Authors:Sarthak Mittal · Eric Elmoznino · Léo Gagnon · Sangnie Bhardwaj · Guillaume Lajoie · Dhanya Sridhar

Title: Does learning the right latent variables necessarily improve in-context learning?

Abstract:
Large autoregressive models like Transformers can solve tasks through incontext learning (ICL) without learning new weights, suggesting avenues for efficiently solving new tasks. For many tasks, e.g., linear regression, the data factorizes: examples are independent given a task latent that generates the data, e.g., linear coefficients. While an optimal predictor leverages this factorization by inferring task latents, it is unclear if Transformers implicitly do so or instead exploit heuristics and statistical shortcuts through attention layers. In this paper, we systematically investigate the effect of explicitly inferring task latents by minimally modifying the Transformer architecture with a bottleneck to prevent shortcuts and incentivize structured solutions. We compare it against standard Transformers across various ICL tasks and find that contrary to intuition and recent works, there is little discernible difference between the two; biasing towards task-relevant latent variables does not lead to better out-of-distribution performance, in general. Curiously, we find that while the bottleneck effectively learns to extract latent task variables from context, downstream processing struggles to utilize them for robust prediction. Our study highlights the intrinsic limitations of Transformers in achieving structured ICL solutions that generalize, and shows that while inferring the right latents aids interpretability, it is not sufficient to alleviate this problem.

Paperid:411

Authors:Fan Nie · Xiaotian Hou · Shuhang Lin · James Zou · Huaxiu Yao · Linjun Zhang

Title: FactTest: Factuality Testing in Large Language Models with Statistical Guarantees

Abstract:
The propensity of large language models (LLMs) to generate hallucinations and nonfactual content undermines their reliability in high-stakes domains, where rigorous control over Type I errors (the conditional probability of incorrectly classifying hallucinations as truthful content) is essential. Despite its importance, formal verification of LLM factuality with such guarantees remains largely unexplored.In this paper, we introduce FactTest, a novel framework that statistically assesses whether an LLM can provide correct answers to given questions with high-probability correctness guarantees. We formulate hallucination detection as a hypothesis testing problem to enforce an upper bound of Type I errors at user-specified significance levels. Notably, we prove that FactTest also ensures strong Type II error control under mild conditions and can be extended to maintain its effectiveness when covariate shifts exist. Our approach is distribution-free and works for any number of human-annotated samples. It is model-agnostic and applies to any black-box or white-box LM. Extensive experiments on question-answering (QA) benchmarks demonstrate that FactTest effectively detects hallucinations and enable LLMs to abstain from answering unknown questions, leading to an over 40% accuracy improvement.

Paperid:412

Authors:Xiuyuan Wang · Chaochao Chen · Weiming Liu · Xinting Liao · Fan Wang · Xiaolin Zheng

Title: Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization

Abstract:
With growing privacy concerns and the enforcement of data protection regulations, machine unlearning has emerged as a promising approach for removing the influence of forget data while maintaining model performance on retain data. However, most existing unlearning methods require access to the original training data, which is often impractical due to privacy policies, storage constraints, and other limitations. This gives rise to the challenging task of sourcefree unlearning, where unlearning must be accomplished without access to the original training data. Few existing source-free unlearning methods rely on knowledge distillation and model retraining, which impose substantial computational costs. In this work, we propose the Data Synthesis-based Discrimination-Aware (DSDA) unlearning framework, which enables efficient source-free unlearning through two stages: (1) Accelerated Energy-Guided Data Synthesis (AEGDS), which employs Langevin dynamics to model the training data distribution while integrating Runge–Kutta methods and momentum to enhance efficiency. (2) Discrimination-Aware Multitask Optimization (DAMO), which refines the feature distribution of retain data and mitigates the gradient conflicts among multiple unlearning objectives. Extensive experiments on three benchmark datasets demonstrate that DSDA outperforms existing unlearning methods, validating its effectiveness and efficiency in source-free unlearning.

Paperid:413

Authors:Minjong Yoo · Jinwoo Jang · Sihyung Yoon · Honguk Woo

Title: World Model Implanting for Test-time Adaptation of Embodied Agents

Abstract:
In embodied AI, a persistent challenge is enabling agents to robustly adapt to novel domains without requiring extensive data collection or retraining. To address this, we present a world model implanting framework (WorMI) that combines the reasoning capabilities of large language models (LLMs) with independently learned, domainspecific world models through test-time composition. By allowing seamless implantation and removal of the world models, the embodied agent's policy achieves and maintains cross-domain adaptability. In the WorMI framework, we employ a prototype-based world model retrieval approach, utilizing efficient trajectory-based abstract representation matching, to incorporate relevant models into test-time composition. We also develop a world-wise compound attention method that not only integrates the knowledge from the retrieved world models but also aligns their intermediate representations with the reasoning model's representation within the agent's policy. This framework design effectively fuses domain-specific knowledge from multiple world models, ensuring robust adaptation to unseen domains. We evaluate our WorMI on the VirtualHome and ALFWorld benchmarks, demonstrating superior zero-shot and few-shot performance compared to several LLM-based approaches across a range of unseen domains. These results highlight the framework’s potential for scalable, real-world deployment in embodied agent scenarios where adaptability and data efficiency are essential.

Paperid:414

Authors:Xiaoyu Yang · Lijian Xu · Hongsheng Li · Shaoting Zhang

Title: One Leaf Reveals the Season: Occlusion-Based Contrastive Learning with Semantic-Aware Views for Efficient Visual Representation

Abstract:
This paper proposes a scalable and straightforward pretraining paradigm for efficient visual conceptual representation called occluded image contrastive learning (OCL). Our OCL approach is simple: we randomly mask patches to generate different views within an image and contrast them among a mini-batch of images. The core idea behind OCL consists of two designs. First, masked tokens have the potential to significantly diminish the conceptual redundancy inherent in images, and create distinct views with substantial fine-grained differences on the semantic concept level instead of the instance level. Second, contrastive learning is adept at extracting high-level semantic conceptual features during the pre-training, circumventing the high-frequency interference and additional costs associated with image reconstruction. Importantly, OCL learns highly semantic conceptual representations efficiently without relying on hand-crafted data augmentations or additional auxiliary modules. Empirically, OCL demonstrates high scalability with Vision Transformers, as the ViT-L/16 can complete pre-training in 133 hours using only 4 A100 GPUs, achieving 85.8\% accuracy in downstream fine-tuning tasks. Code is available at https://anonymous.4open.science/r/OLRS/.

Paperid:415

Authors:Bin Luo · Yuwen Huang · Jonathan Allcock · Xiaojun Lin · Shengyu Zhang · John C. S. Lui

Title: Quantum Algorithms for Finite-horizon Markov Decision Processes

Abstract:
Abstract:In this work, we design quantum algorithms that are more efficient than classical algorithms to solve timedependent and finite-horizon Markov Decision Processes (MDPs) in two distinct settings: (1) In the exact dynamics setting, where the agent has full knowledge of the environment's dynamics (i.e., transition probabilities), we prove that our algorithm **QVI-1** achieves a quadratic speedup in the size of the action space $(A)$ compared with the classical value iteration algorithm for computing the optimal policy ($\pi^{\*}$) and the optimal V-value function ($V_{0}^{\*}$). Furthermore, our algorithm **QVI-2** provides an additional speedup in the size of the state space $(S)$ when obtaining near-optimal policies and V-value functions. (2) In the generative model setting, where samples from the environment are accessible in quantum superposition, we prove that our algorithms **QVI-3** and **QVI-4** achieve improvements in sample complexity over the state-of-the-art (SOTA) classical algorithm in terms of $A$, estimation error $(\epsilon)$, and time horizon $(H)$. More importantly, we prove quantum lower bounds to show that **QVI-3** and **QVI-4** are nearly minimax optimal.

Paperid:416

Authors:Rina Bao · Shilong Dong · Zhenfang Chen · Sheng He · Patricia Ellen Grant · Yangming Ou

Title: Visual and Domain Knowledge for Professional-level Graph-of-Thought Medical Reasoning

Abstract:
Medical Visual Question Answering (MVQA) requires AI models to answer questions related to medical images, offering significant potential to assist medical professionals in evaluating and diagnosing diseases, thereby improving early interventions. However, existing MVQA datasets primarily focus on basic questions regarding visual perception and pattern recognition, without addressing the more complex questions that are critical in clinical diagnosis and decisionmaking. This paper introduces a new dataset designed for professional-level medical reasoning, simulating the decision-making process. We achieve this by collecting MRI and clinical data related to Hypoxic-Ischemic Encephalopathy (HIE, a neonatal brain injury that occurs in 1-5/1000 neonates) over a decade of effort, enriched with expert annotations and insights. Building on this data, we generate clinical question-answer pairs and MRI interpretations to enable comprehensive diagnosis, interpretation, and prediction of neurocognitive outcomes. Our evaluation of current large vision-language models (LVLMs) shows limited performance on this benchmark, highlighting both the challenges of the task and the importance of this benchmark for advancing medical AI. Furthermore, we propose a novel ``Clinical Graph of Thoughts" (CGoT) model, which integrates domain-specific medical knowledge and clinical reasoning processes with the interpretive abilities of LVLMs. The model demonstrates promising results, achieving around 15% absolute gain on the most important neurocognitive outcome task, while the dataset still reveals substantial opportunities for further research innovation.

Paperid:417

Authors:Zhenyu Hou · Xin Lv · Rui Lu · Jiajie Zhang · Yujiang Li · Zijun Yao · Juanzi Li · Jie Tang · Yuxiao Dong

Title: Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling

Abstract:
Large language models (LLMs) have demonstrated remarkable capabilities in complex reasoning tasks. However, existing approaches mainly rely on imitation learning and struggle to achieve effective testtime scaling. While reinforcement learning (RL) holds promise for enabling self-exploration, recent attempts yield modest improvements in complex reasoning. In this paper, we present T1 to scale RL by encouraging exploration and understand inference scaling. We first initialize the LLM using synthesized chain-of-thought data that integrates trial-and-error and self-verification. To scale RL training, we promote increased sampling diversity through oversampling. We demonstrate that T1 with open LLMs as its base exhibits inference scaling behavior and achieves superior performance on challenging math reasoning benchmarks. More importantly, we present a simple strategy to examine inference scaling, where increased inference budgets directly lead to T1's better performance without any additional verification.

Paperid:418

Authors:Herman Chau · Helen Jenne · Davis Brown · Jesse He · Mark Raugas · Sara Billey · Henry Kvinge

Title: Machine Learning meets Algebraic Combinatorics: A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics

Abstract:
With recent dramatic increases in AI system capabilities, there has been growing interest in utilizing machine learning for reasoningheavy, quantitative tasks, particularly mathematics. While there are many resources capturing mathematics at the high-school, undergraduate, and graduate level, there are far fewer resources available that align with the level of difficulty and open endedness encountered by professional mathematicians working on open problems. To address this, we introduce a new collection of datasets, the Algebraic Combinatorics Dataset Repository (ACD Repo), representing either foundational results or open problems in algebraic combinatorics, a subfield of mathematics that studies discrete structures arising from abstract algebra. Further differentiating our dataset collection is the fact that it aims at the conjecturing process. Each dataset includes a research level question and then a large collection of examples (up to 10M in some cases) from which conjectures should be generated. We describe all nine datasets, the different ways machine learning models can be applied to them (e.g., training with narrow models followed by mechanistic interpretability or program synthesis with LLMs), and document some examples where AI systems were able to re-discover non-trivial known results.

Paperid:419

Authors:Andrew Wagenmaker · Zhiyuan Zhou · Sergey Levine

Title: Training Your Agent to Explore via In-Context Adaptation

Abstract:
Developing autonomous agents that quickly explore an environment and adapt their behavior online is a canonical challenge in robotics and machine learning. While humans are able to achieve such fast online exploration and adaptation, often acquiring new information and skills in only a handful of interactions, existing algorithmic approaches tend to rely on random exploration and slow, gradientbased behavior updates. How can we endow autonomous agents with such capabilities on par with humans? Taking inspiration from recent progress on both in-context learning and large-scale behavioral cloning, in this work we propose training agents to internalize what it means to explore and adapt in-context over the space of ''expert'' behaviors. To achieve this, given access to a dataset of expert demonstrations, we train a long-context generative model to predict expert actions conditioned on a context of past observations and a measure of how ''exploratory'' the expert's behaviors are relative to this context. This enables the model to not only mimic the behavior of an expert, but also, by feeding its past history of interactions into its context, to select different expert behaviors than what have been previously selected, thereby allowing for fast online adaptation and targeted, ''expert-like'' exploration. We demonstrate the effectiveness of our method in both simulated locomotion and manipulation settings, as well as on real-world robotic manipulation tasks, illustrating its ability to learn adaptive, exploratory behavior.

Paperid:420

Authors:Xinzhi Zhang · Hohei Chan · Deheng Ye · Yi Cai · Mengchen Zhao

Title: Ad Hoc Teamwork via Offline Goal-Based Decision Transformers

Abstract:
The ability of agents to collaborate with previously unknown teammates on the fly, known as ad hoc teamwork (AHT), is crucial in many realworld applications. Existing approaches to AHT require online interactions with the environment and some carefully designed teammates. However, these prerequisites can be infeasible in practice. In this work, we extend the AHT problem to the offline setting, where the policy of the ego agent is directly learned from a multi-agent interaction dataset. We propose a hierarchical sequence modeling framework called TAGET that addresses critical challenges in the offline setting, including limited data, partial observability and online adaptation. The core idea of TAGET is to dynamically predict teammate-aware rewards-to-go and sub-goals, so that the ego agent can adapt to the changes of teammates' behaviors in real time. Extensive experimental results show that TAGET significantly outperforms existing solutions to AHT in the offline setting.

Paperid:421

Authors:Guoqing Zhang · Shichao Kan · Fanghui Zhang · Wanru Xu · Yue Zhang · Yigang Cen

Title: Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization for Scene Graph Generation

Abstract:
Scene Graph Generation (SGG) is a fundamental task in visual understanding, aimed at providing more precise local detail comprehension for downstream applications. Existing SGG methods often overlook the diversity of predicate representations and the consistency among similar predicates when dealing with longtail distributions. As a result, the model's decision layer fails to effectively capture details from the tail end, leading to biased predictions. To address this, we propose a Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization (NoDIS) method. On the one hand, expanding the predicate representation space enhances the model's ability to learn both common and rare predicates, thus reducing prediction bias caused by data scarcity. We propose a conditional diffusion model to reconstructs features and increase the diversity of representations for same category predicates. On the other hand, independent predicate representations in the decision phase increase the learning complexity of the decision layer, making accurate predictions more challenging. To address this issue, we introduce a discretization mapper that learns consistent representations among similar predicates, reducing the learning difficulty and decision ambiguity in the decision layer. To validate the effectiveness of our method, we integrate NoDIS with various SGG baseline models and conduct experiments on multiple datasets. The results consistently demonstrate superior performance.

Paperid:422

Authors:Kajetan Schweighofer · Adrián Arnaiz-Rodríguez · Sepp Hochreiter · Nuria Oliver

Title: The Disparate Benefits of Deep Ensembles

Abstract:
Ensembles of Deep Neural Networks, Deep Ensembles, are widely used as a simple way to boost predictive performance. However, their impact on algorithmic fairness is not well understood yet. Algorithmic fairness examines how a model's performance varies across socially relevant groups defined by protected attributes such as age, gender, or race. In this work, we explore the interplay between the performance gains from Deep Ensembles and fairness. Our analysis reveals that they unevenly favor different groups, a phenomenon we term the disparate benefits effect. We empirically investigate this effect using popular facial analysis and medical imaging datasets with protected group attributes and find that it affects multiple established group fairness metrics, including statistical parity and equal opportunity. Furthermore, we identify that the pergroup differences in predictive diversity of ensemble members can explain this effect. Finally, we demonstrate that the classical Hardt post-processing method is particularly effective at mitigating the disparate benefits effect of Deep Ensembles by leveraging their better-calibrated predictive distributions.

Paperid:423

Authors:Chunming He · Rihan Zhang · Fengyang Xiao · Chengyu Fang · Longxiang Tang · Yulun Zhang · Linghe Kong · Deng-Ping Fan · Kai Li · Sina Farsiu

Title: RUN: Reversible Unfolding Network for Concealed Object Segmentation

Abstract:
Concealed object segmentation (COS) is a challenging problem that focuses on identifying objects that are visually blended into their background. Existing methods often employ reversible strategies to concentrate on uncertain regions but only focus on the mask level, overlooking the valuable of the RGB domain. To address this, we propose a Reversible Unfolding Network (RUN) in this paper. RUN formulates the COS task as a foregroundbackground separation process and incorporates an extra residual sparsity constraint to minimize segmentation uncertainties. The optimization solution of the proposed model is unfolded into a multistage network, allowing the original fixed parameters to become learnable. Each stage of RUN consists of two reversible modules: the Segmentation-Oriented Foreground Separation (SOFS) module and the Reconstruction-Oriented Background Extraction (ROBE) module. SOFS applies the reversible strategy at the mask level and introduces Reversible State Space to capture non-local information. ROBE extends this to the RGB domain, employing a reconstruction network to address conflicting foreground and background regions identified as distortion-prone areas, which arise from their separate estimation by independent modules. As the stages progress, RUN gradually facilitates reversible modeling of foreground and background in both the mask and RGB domains, reducing false-positive and false-negative regions. Extensive experiments demonstrate the superior performance of RUN and underscore the promise of unfolding-based frameworks for COS and other high-level vision tasks. The source code will be made publicly available.

Paperid:424

Authors:Richeng Jin · Huaiyu Dai

Title: Noisy SIGNSGD Is More Differentially Private Than You (Might) Think

Abstract:
The prevalent distributed machine learning paradigm faces two critical challenges: communication efficiency and data privacy. SIGNSGD provides a simpleto-implement approach with improved communication efficiency by requiring workers to share only the signs of the gradients. However, it fails to converge in the presence of data heterogeneity, and a simple fix is to add Gaussian noise before taking the signs, which leads to the Noisy SIGNSGD algorithm that enjoys competitive performance while significantly reducing the communication overhead. Existing results suggest that Noisy SIGNSGD with additive Gaussian noise has the same privacy guarantee as classic DP-SGD due to the post-processing property of differential privacy, and logistic noise may be a good alternative to Gaussian noise when combined with the sign-based compressor. Nonetheless, discarding the magnitudes in Noisy SIGNSGD leads to information loss, which may intuitively amplify privacy. In this paper, we make this intuition rigorous and quantify the privacy amplification of the sign-based compressor. Particularly, we analytically show that Gaussian noise leads to a smaller estimation error than logistic noise when combined with the sign-based compressor and may be more suitable for distributed learning with heterogeneous data. Then, we further establish the convergence of Noisy SIGNSGD. Finally, extensive experiments are conducted to validate the theoretical results.

Paperid:425

Authors:Huigen Ye · Hua Xu · An Yan · Yaoyang Cheng

Title: Large Language Model-driven Large Neighborhood Search for Large-Scale MILP Problems

Abstract:
Large Neighborhood Search (LNS) is a widely used method for solving largescale Mixed Integer Linear Programming (MILP) problems. The effectiveness of LNS crucially depends on the choice of the search neighborhood. However, existing strategies either rely on expert knowledge or computationally expensive Machine Learning (ML) approaches, both of which struggle to scale effectively for large problems. To address this, we propose LLM-LNS, a novel Large Language Model (LLM)-driven LNS framework for large-scale MILP problems. Our approach introduces a dual-layer self-evolutionary LLM agent to automate neighborhood selection, discovering effective strategies with scant small-scale training data that generalize well to large-scale MILPs. The inner layer evolves heuristic strategies to ensure convergence, while the outer layer evolves evolutionary prompt strategies to maintain diversity. Experimental results demonstrate that the proposed dual-layer agent outperforms state-of-the-art agents such as FunSearch and EOH. Furthermore, the full LLM-LNS framework surpasses manually designed LNS algorithms like ACP, ML-based LNS methods like CL-LNS, and large-scale solvers such as Gurobi and SCIP. It also achieves superior performance compared to advanced ML-based MILP optimization frameworks like GNN&GBDT and Light-MILPopt, further validating the effectiveness of our approach.

Paperid:426

Authors:Jinpeng Chen · Runmin Cong · Yuzhi Zhao · Hongzheng Yang · Guangneng Hu · Horace Ip · Sam Kwong

Title: SEFE: Superficial and Essential Forgetting Eliminator for Multimodal Continual Instruction Tuning

Abstract:
Multimodal Continual Instruction Tuning (MCIT) aims to enable Multimodal Large Language Models (MLLMs) to incrementally learn new tasks without catastrophic forgetting, thus adapting to evolving requirements. In this paper, we explore the forgetting caused by such incremental training, categorizing it into superficial forgetting and essential forgetting. Superficial forgetting refers to cases where the model’s knowledge may not be genuinely lost, but its responses to previous tasks deviate from expected formats due to the influence of subsequent tasks’ answer styles, making the results unusable. On the other hand, essential forgetting refers to situations where the model provides correctly formatted but factually inaccurate answers, indicating a true loss of knowledge. Assessing essential forgetting necessitates addressing superficial forgetting first, as severe superficial forgetting can conceal the model’s knowledge state. Hence, we first introduce the Answer Style Diversification (ASD) paradigm, which defines a standardized process for data style transformations across different tasks, unifying their training sets into similarly diversified styles to prevent superficial forgetting caused by style shifts. Building on this, we propose RegLoRA to mitigate essential forgetting. RegLoRA stabilizes key parameters where prior knowledge is primarily stored by applying regularization to LoRA’s weight update matrices, enabling the model to retain existing competencies while remaining adaptable to new tasks. Experimental results demonstrate that our overall method, SEFE, achieves stateof-the-art performance.

Paperid:427

Authors:Xiaomin Li · Mingye Gao · Zhiwei Zhang · Jingxuan Fan · Weiyu Li

Title: RuleAdapter: Dynamic Rules for training Safety Reward Models in RLHF

Abstract:
Reinforcement Learning from Human Feedback (RLHF) is widely used to align models with human preferences, particularly to enhance the safety of responses generated by LLMs. This method traditionally relies on choosing preferred responses from response pairs. However, due to variations in human opinions and the difficulty of making an overall comparison of two responses, there is a growing shift towards a finegrained annotation approach, assessing responses based on multiple specific metrics or rules. Selecting and applying these rules efficiently while accommodating the diversity of preference data remains a significant challenge. In this paper, we introduce a dynamic approach that adaptively selects the most critical rules for each pair of responses. We develop a mathematical framework that leverages the maximum discrepancy between each paired responses and theoretically show that this strategy optimizes the mutual information between the rule-based labeling and the hidden ground-truth preferences. We then train an 8B reward model using the adaptively labeled preference dataset and evaluate its performance on RewardBench. As of January 25, 2025, our model achieved the highest safety performance on the leaderboard, outperforming various larger models.

Paperid:428

Authors:Abudukelimu Wuerkaixi · Qizhou Wang · Sen Cui · Wutong Xu · Bo Han · Gang Niu · Masashi Sugiyama · Changshui Zhang

Title: Adaptive Localization of Knowledge Negation for Continual LLM Unlearning

Abstract:
With the growing deployment of large language models (LLMs) across diverse domains, concerns regarding their safety have grown substantially.LLM unlearning has emerged as a pivotal approach to removing harmful or unlawful contents while maintaining utility.Despite increasing interest, the challenges of continual unlearning, which is common in realworld scenarios, remain underexplored.Successive unlearning tasks often lead to intensified utility degradation.To effectively unlearn targeted knowledge while preserving LLM utility, it is essential to minimize changes in model parameters by selectively updating those linked to the target knowledge, thereby ensuring other knowledge remains unaffected.Building on the task vector framework, we propose a new method named ALKN (Adaptive Localization of Knowledge Negation), which uses dynamic masking to sparsify training gradients and adaptively adjusts unlearning intensity based on inter-task relationships.Comprehensive experiments across three well-established LLM unlearning datasets demonstrate that our approach consistently outperforms baseline methods in both unlearning effectiveness and utility retention under continual unlearning settings.

Paperid:429

Authors:Ari Karchmer · Eran Malach

Title: The Power of Random Features and the Limits of Distribution-Free Gradient Descent

Abstract:
We study the relationship between gradientbased optimization of parametric models (e.g., neural networks) and optimization of linear combinations of random features. Our main result shows that if a parametric model can be learned using mini-batch stochastic gradient descent (bSGD) without making assumptions about the data distribution, then with high probability, the target function can also be approximated using a polynomial-sized combination of random features. The size of this combination depends on the number of gradient steps and numerical precision used in the bSGD process. This finding reveals fundamental limitations of distribution-free learning in neural networks trained by gradient descent, highlighting why making assumptions about data distributions is often crucial in practice.Along the way, we also introduce a new theoretical framework called average probabilistic dimension complexity (adc), which extends the probabilistic dimension complexity developed by Kamath et al. (2020). We prove that adc has a polynomial relationship with statistical query dimension, and use this relationship to demonstrate an infinite separation between adc and standard dimension complexity.

Paperid:430

Authors:Hoigi Seo · Wongi Jeong · Jae-sun Seo · Se Young Chun

Title: Skrr: Skip and Re-use Text Encoder Layers for Memory Efficient Text-to-Image Generation

Abstract:
Largescale text encoders in text-to-image (T2I) diffusion models have demonstrated exceptional performance in generating high-quality images from textual prompts. Unlike denoising modules that rely on multiple iterative steps, text encoders require only a single forward pass to produce text embeddings. However, despite their minimal contribution to total inference time and floating-point operations (FLOPs), text encoders demand significantly higher memory usage, up to eight times more than denoising modules. To address this inefficiency, we propose Skip and Re-use layers (Skrr), a simple yet effective pruning strategy specifically designed for text encoders in T2I diffusion models. Skrr exploits the inherent redundancy in transformer blocks by selectively skipping or reusing certain layers in a manner tailored for T2I tasks, thereby reducing memory consumption without compromising performance. Extensive experiments demonstrate that Skrr maintains image quality comparable to the original model even under high sparsity levels, outperforming existing blockwise pruning methods. Furthermore, Skrr achieves state-of-the-art memory efficiency while preserving performance across multiple evaluation metrics, including the FID, CLIP, DreamSim, and GenEval scores.

Paperid:431

Authors:Hongyang Lei · Xiaolong Cheng · Dan Wang · Qi Qin · Huazhen Huang · Yetao Wu · Qingqing Gu · Luo Ji

Title: M3-Jepa: Multimodal Alignment via Multi-directional MoE based on the JEPA framework

Abstract:
Current multimodal alignment strategies primarily use single or unified modality encoders, while optimizing the alignment on the original token space. Such a framework is easy to implement and incorporate with the pretrained knowledge, but might result in information bias. To deal with such issues, the joint encoding predictive architecture (JEPA) learns the alignment loss on the latent space, with a predictor to convert the input encoding to the output latent space. However, the application of JEPA in multimodal scenarios is limited so far. In this paper, we introduce M3Jepa, a scalable multimodal alignment framework, with the predictor implemented by a multi-directional mixture of experts (MoE). We demonstrate the framework can maximize the mutual information with information theory derivations, by alternating the optimization between different uni-directional tasks. By thoroughly designed experiments, we show that M3-Jepa can obtain state-of-the-art performance on different modalities and tasks, generalize to unseen datasets and domains, and is computationally efficient in training and inference. Our study indicates that M3-Jepa might provide a new paradigm to self-supervised learning and open-world modeling.

Paperid:432

Authors:Jinlong Pang · Na Di · Zhaowei Zhu · Jiaheng Wei · Hao Cheng · Chen Qian · Yang Liu

Title: Token Cleaning: Fine-Grained Data Selection for LLM Supervised Fine-Tuning

Abstract:
Recent studies show that in supervised finetuning (SFT) of large language models (LLMs), data quality matters more than quantity. While most data cleaning methods concentrate on filtering entire samples, the quality of individual tokens within a sample can vary significantly. After pre-training, even in high-quality samples, patterns or phrases that are not task-related can be redundant or uninformative. Continuing to fine-tune on these patterns may offer limited benefit and even degrade downstream task performance. In this paper, we investigate token quality from a noisy-label perspective and propose a generic token cleaning pipeline for SFT tasks. Our method filters out uninformative tokens while preserving those carrying key task-specific information. Specifically, we first evaluate token quality by examining the influence of model updates on each token, then apply a threshold-based separation. The token influence can be measured in a single pass with a fixed reference model or iteratively with self-evolving reference models. The benefits and limitations of both methods are analyzed theoretically by error upper bounds. Extensive experiments show that our framework consistently improves performance across multiple downstream tasks.

Paperid:433

Authors:Shuoyuan Wang · Sharon Li · Hongxin Wei

Title: Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models

Abstract:
Confidence calibration is critical for the safe deployment of machine learning models in the real world. However, such issue in visionlanguage models like CLIP, particularly after fine-tuning, has not been fully addressed. In this work, we demonstrate that existing prompt tuning methods usually lead to a trade-off of calibration between base and new classes: the cross-entropy loss used in standard fine-tuning (e.g., CoOp) causes overconfidence in new classes by increasing textual label divergence, whereas regularization-based tuning (e.g., KgCoOp) maintains the confidence level but results in underconfidence in base classes due to the improved accuracy. Inspired by the observations, we introduce Dynamic Outlier Regularization (DOR) to ensure the confidence calibration on both base and new classes after fine-tuning. In particular, we propose to minimize the feature deviation of novel textual labels (instead of base classes) sampled from a large vocabulary. In effect, DOR prevents the increase in textual divergence for new labels while easing restrictions on base classes. Extensive experiments demonstrate that DOR can enhance the calibration performance of current fine-tuning methods on base and new classes.

Paperid:434

Authors:Shion Takeno · Yoshito Okura · Yu Inatsu · Tatsuya Aoyama · Tomonari Tanaka · Satoshi Akahane · Hiroyuki Hanada · Noriaki Hashimoto · Taro Murayama · Hanju Lee · Shinya Kojima · Ichiro Takeuchi

Title: Distributionally Robust Active Learning for Gaussian Process Regression

Abstract:
Gaussian process regression (GPR) or kernel ridge regression is a widely used and powerful tool for nonlinear prediction. Therefore, active learning (AL) for GPR, which actively collects data labels to achieve an accurate prediction with fewer data labels, is an important problem. However, existing AL methods do not theoretically guarantee prediction accuracy for target distribution. Furthermore, as discussed in the distributionally robust learning literature, specifying the target distribution is often difficult. Thus, this paper proposes two AL methods that effectively reduce the worstcase expected error for GPR, which is the worst-case expectation in target distribution candidates. We show an upper bound of the worst-case expected squared error, which suggests that the error will be arbitrarily small by a finite number of data labels under mild conditions. Finally, we demonstrate the effectiveness of the proposed methods through synthetic and real-world datasets.

Paperid:435

Authors:Pranjal Aggarwal · Bryan Parno · Sean Welleck

Title: AlphaVerus: Bootstrapping Formally Verified Code Generation through Self-Improving Translation and Treefinement

Abstract:
Automated code generation with large language models has gained significant traction, but there remains no guarantee of the correctness of generated code. We aim to use formal verification to provide mathematical guarantees that the generated code is correct. However, generating formally verified code with LLMs is hindered by the scarcity of training data and the complexity of formal proofs. To tackle this challenge, we introduce AlphaVerus, a selfimproving framework that bootstraps formally verified code generation by iteratively translating programs from a higher-resource language and leveraging feedback from a verifier. AlphaVerus operates in three phases: exploration of candidate translations, Treefinement -- a novel tree search algorithm for program refinement using verifier feedback, and filtering misaligned specifications and programs to prevent reward hacking. Through this iterative process, AlphaVerus enables LLaMA-3.1-70B model to generate verified code without human intervention or model finetuning. AlphaVerus shows an ability to generate formally verified solutions for HumanEval and MBPP, laying the groundwork for truly trustworthy code-generation agents.

Paperid:436

Authors:KaShun SHUM · Yuzhen Huang · Hongjian Zou · dingqi · YiXuan Liao · Xiaoxin Chen · Qian Liu · Junxian He

Title: Predictive Data Selection: The Data That Predicts Is the Data That Teaches

Abstract:
Language model pretraining involves training on extensive corpora, where data quality plays a pivotal role. In this work, we aim to directly estimate the contribution of data during pretraining and select pretraining data in an efficient manner. Specifically, we draw inspiration from recent findings showing that compression efficiency (i.e., the normalized loss) of diverse models on certain text correlates strongly with their downstream performance, when the text domain aligns with the downstream benchmark (Huang et al., 2024). Building on this observation, we hypothesize thatdata on which model losses are predictive of downstream abilities also contribute effectively to learning. To leverage this insight, we introduce data selection based on data's Predictive strength (Preselect), a lightweight and efficient data selection method that requires training and deploying only a fastTextbased scorer. Through comprehensive experiments with 1B and 3B parameter models, we demonstrate that PreSelect trained on 30B tokens surpasses the performance of a vanilla baseline trained on 300B tokens, achieving a 10x reduction in compute requirements. Furthermore, PreSelect significantly outperforms other competitive data selection baselines, such as DCLM and FineWeb-Edu. To facilitate further research, we will open-source our trained data selection scorer along with the curated datasets for the community.

Paperid:437

Authors:Yaoqin He · Junchen Fu · Kaiwen Zheng · Songpei Xu · Fuhai Chen · Jie Li · Joemon Jose · Xuri Ge

Title: Double-Filter: Efficient Fine-tuning of Pre-trained Vision-Language Models via Patch&Layer Filtering

Abstract:
In this paper, we present a novel approach, termed DoubleFilter, to ``slim down'' the fine-tuning process of vision-language pre-trained (VLP) models via filtering redundancies in feature inputs and architectural components. We enhance the fine-tuning process using two approaches. First, we develop a new patch selection method incorporating image patch filtering through background and foreground separation, followed by a refined patch selection process. Second, we design a genetic algorithm to eliminate redundant fine-grained architecture layers, improving the efficiency and effectiveness of the model. The former makes patch selection semantics more comprehensive, improving inference efficiency while ensuring semantic representation. The latter's fine-grained layer filter removes architectural redundancy to the extent possible and mitigates the impact on performance. Experimental results demonstrate that the proposed Double-Filter achieves superior efficiency of model fine-tuning and maintains competitive performance compared with the advanced efficient fine-tuning methods on two downstream tasks, VQA and NLVR. In addition, it has been proven to be effective under METER and ViLT VLP models.

Paperid:438

Authors:Xilin Wei · Xiaoran Liu · Yuhang Zang · Xiaoyi Dong · Pan Zhang · Yuhang Cao · Jian Tong · Haodong Duan · Qipeng Guo · Jiaqi Wang · Xipeng Qiu · Dahua Lin

Title: VideoRoPE: What Makes for Good Video Rotary Position Embedding?

Abstract:
While Rotary Position Embedding (RoPE) and its variants are widely adopted for their longcontext capabilities, the extension of the 1D RoPE to video, with its complex spatio-temporal structure, remains an open challenge.This work first introduces a comprehensive analysis that identifies four key characteristics essential for the effective adaptation of RoPE to video, which have not been fully considered in prior work.As part of our analysis, we introduce a challenging V-NIAH-D (Visual Needle-In-A-Haystack with Distractors) task, which adds periodic distractors into V-NIAH.The V-NIAH-D task demonstrates that previous RoPE variants, lacking appropriate temporal dimension allocation, are easily misled by distractors.Based on our analysis, we introduce VideoRoPE, with a 3D structure designed to preserve spatio-temporal relationships.VideoRoPE features low-frequency temporal allocation to mitigate periodic oscillations, a diagonal layout to maintain spatial symmetry, and adjustable temporal spacing to decouple temporal and spatial indexing.VideoRoPE consistently surpasses previous RoPE variants, across diverse downstream tasks such as long video retrieval, video understanding, and video hallucination.Our code and model weights will be publicly released.

Paperid:439

Authors:Nikolaus Howe · Ian McKenzie · Oskar Hollinsworth · Michał Zając · Tom Tseng · Aaron Tucker · Pierre-Luc Bacon · Adam Gleave

Title: Scaling Trends in Language Model Robustness

Abstract:
Language models exhibit scaling laws, whereby increasing model and dataset size predictably decreases negative log likelihood, unlocking a dazzling array of capabilities. At the same time, even the most capable systems are currently vulnerable to adversarial inputs such as jailbreaks and prompt injections, despite concerted efforts to make them robust. As compute becomes more accessible to both attackers and defenders, which side will benefit more from scale?We attempt to answer this question with a detailed study of robustness on language models spanning three orders of magnitude in parameter count. From the defender's perspective, we find that in the absence of other interventions, increasing model size alone does not consistently improve robustness. In adversarial training, we find that larger models are more sampleefficient and less compute-efficient than smaller models, and often better generalize their defense to new threat models. From the attacker’s perspective, we find that increasing attack compute smoothly and reliably increases attack success rate against both finetuned and adversarially trained models. Finally, we show that across model sizes studied, doubling compute on adversarial training only forces an attacker to less than double attack compute to maintain the same attack success rate. However, adversarial training becomes more and more effective on larger models, suggesting that defenders could eventually have the advantage with increasing model size.These results underscore the value of adopting a scaling lens when discussing robustness of frontier models.

Paperid:440

Authors:Mostafa Elhoushi · Jeff Johnson

Title: any4: Learned 4-bit Numeric Representation for LLMs

Abstract:
We present any4, a learned 4bit weight quantization solution for large language models (LLMs) providing arbitrary numeric representations without requiring pre-processing of weights or activations. any4 yields higher accuracy compared to other related 4-bit numeric representation types: int4, fp4 and nf4, as evaluated on a range of model sizes, generations and families (Llama 2, Llama 3, Mistral and Mixtral). While any4 does not require preprocessing of weights or activations, it is also competitive with orthogonal techniques that require such preprocessing (e.g., AWQ and GPTQ). We also experiment with any3 and any2 and show competitiveness at lower bits. Additionally, we show that we can calibrate using a single curated diverse sample rather than hundreds of samples from a dataset as done in most quantization approaches. We also open source tinygemm, a latency optimized GPU matrix multiplication library for LLMs, that implements any4 using a GPU-efficient lookup table strategy along with other common quantization methods.

Paperid:441

Authors:Hongyuan Su · Yu Zheng · Yuan Yuan · Yuming Lin · Depeng Jin · Yong Li

Title: Reinforcement Learning with Adaptive Reward Modeling for Expensive-to-Evaluate Systems

Abstract:
Training reinforcement learning (RL) agents requires extensive trials and errors, which becomes prohibitively timeconsuming in systems with costly reward evaluations.To address this challenge, we propose adaptive reward modeling (AdaReMo) which accelerates RL training by decomposing the complicated reward function into multiple localized fast reward models approximating direct reward evaluation with neural networks.These models dynamically adapt to the agent’s evolving policy by fitting the currently explored subspace with the latest trajectories, ensuring accurate reward estimation throughout the entire training process while significantly reducing computational overhead.We empirically show that AdaReMo not only achieves over 1,000 times speedup but also improves the performance by 14.6% over state-of-the-art approaches across three expensive-to-evaluate systems---molecular generation, epidemic control, and spatial planning.Code and data for the project are provided at https://anonymous.4open.science/r/RLS-CFCE

Paperid:442

Authors:Run He · Di Fang · Yicheng Xu · Yawen Cui · Ming Li · Cen Chen · Ziqian Zeng · HUIPING ZHUANG

Title: Semantic Shift Estimation via Dual-Projection and Classifier Reconstruction for Exemplar-Free Class-Incremental Learning

Abstract:
ExemplarFree Class-Incremental Learning (EFCIL) aims to sequentially learn from distinct categories without retaining exemplars but easily suffers from catastrophic forgetting of learned knowledge. While existing EFCIL methods leverage knowledge distillation to alleviate forgetting, they still face two critical challenges: semantic shift and decision bias. Specifically, the embeddings of old tasks shift in the embedding space after learning new tasks, and the classifier becomes biased towards new tasks due to training solely with new data, thereby hindering the balance between old and new knowledge. To address these issues, we propose the Dual-Projection Shift Estimation and Classifier Reconstruction (DPCR) approach for EFCIL. DPCR effectively estimates semantic shift through a dual-projection, which combines a learnable transformation with a row-space projection to capture both task-wise and category-wise shifts. Furthermore, to mitigate decision bias, DPCR employs ridge regression to reformulate classifier training as a reconstruction process. This reconstruction exploits previous information encoded in covariance and prototype of each class after calibration with estimated shift, thereby reducing decision bias. Extensive experiments demonstrate that, across various datasets, DPCR effectively balances old and new tasks, outperforming state-of-the-art EFCIL methods.

Paperid:443

Authors:Dujian Ding · Ankur Mallick · Shaokun Zhang · Chi Wang · Daniel Madrigal · Mirian Hipolito Garcia · Menglin Xia · Laks V.S Lakshmanan · Qingyun Wu · Victor Ruehle

Title: BEST-Route: Adaptive LLM Routing with Test-Time Optimal Compute

Abstract:
Large language models (LLMs) are powerful tools but are often expensive to deploy at scale. LLM query routing mitigates this by dynamically assigning queries to models of varying cost and quality to obtain a desired tradeoff. Prior query routing approaches generate only one response from the selected model and a single response from a small (inexpensive) model was often not good enough to beat a response from a large (expensive) model due to which they end up overusing the large model and missing out on potential cost savings. However, it is well known that for small models, generating multiple responses and selecting the best can enhance quality while remaining cheaper than a single largemodel response. We leverage this idea to propose BEST-Route, a novel routing framework that chooses a model and the number of responses to sample from it based on query difficulty and the quality thresholds. Experiments on real-world datasets demonstrate that our method reduces costs by up to 60\% with less than 1\% performance drop.

Paperid:444

Authors:Grant Forbes · Leonardo Villalobos-Arias · Jianxun Wang · Arnav Jhala · David Roberts

Title: Action-Dependent Optimality-Preserving Reward Shaping

Abstract:
Recent RL research has utilized reward shapingparticularly complex shaping rewards such as intrinsic motivation (IM)--to encourage agent exploration in sparse-reward environments. While often effective, ``reward hacking'' can lead to the shaping reward being optimized at the expense of the extrinsic reward, resulting in a suboptimal policy. Potential-Based Reward Shaping (PBRS) techniques such as Generalized Reward Matching (GRM) and Policy-Invariant Explicit Shaping (PIES) have mitigated this. These methods allow for implementing IM without altering optimal policies. In this work we show that they are effectively unsuitable for complex, exploration-heavy environments with long-duration episodes. To remedy this, we introduce Action-Dependent Optimality Preserving Shaping (ADOPS), a method of converting intrinsic rewards to an optimality-preserving form that allows agents to utilize IM more effectively in the extremely sparse environment of Montezuma's Revenge. We also prove ADOPS accommodates reward shaping functions that cannot be written in a potential-based form: while PBRS-based methods require the cumulative discounted intrinsic return be independent of actions, ADOPS allows for intrinsic cumulative returns to be dependent on agents' actions while still preserving the optimal policy set. We show how action-dependence enables ADOPS's to preserve optimality while learning in complex, sparse-reward environments where other methods struggle.

Paperid:445

Authors:Qihong Song · XitingLiu · Hongyuan Zhu · Joey Tianyi Zhou · Xi Peng · Peng Hu

Title: Deep Unsupervised Hashing via External Guidance

Abstract:
Recently, deep unsupervised hashing has gained considerable attention in image retrieval due to its advantages in costfree data labeling, computational efficiency, and storage savings. Although existing methods achieve promising performance by leveraging inherent visual structures within data, they primarily focus on learning discriminative features from unlabeled images through the limited internal knowledge, resulting in an intrinsic upper bound on performance. To break through this intrinsic limitation, we propose a novel method called Deep Unsupervised Hashing with External Guidance (DUH-EG), which incorporates external textual knowledge as semantic guidance to enhance discrete representation learning. Specifically, our DUH-EG: i) first selects representative semantic nouns from an external textual database by minimizing their redundancy, then matches images with them to extract more discriminative external features; and ii) presents a novel bidirectional contrastive learning mechanism to maximize agreement between hash codes in internal and external spaces, thereby capturing discrimination from both external and intrinsic structures in Hamming space. Extensive experiments on four benchmark datasets demonstrate that our DUH-EG remarkably outperforms existing state-of-the-art hashing methods.

Paperid:446

Authors:Lei Yan · Xin Zhang · Qing Mai

Title: Heterogeneous Sufficient Dimension Reduction and Subspace Clustering

Abstract:
Scientific and engineering applications are often heterogeneous, making it beneficial to account for latent clusters or subpopulations when learning low-dimensional subspaces in supervised learning, and vice versa. In this paper, we combine the concept of subspace clustering with model-based sufficient dimension reduction and thus generalize the sufficient dimension reduction framework from homogeneous regression setting to heterogeneous data applications. In particular, we propose the mixture of principal fitted components (mixPFC) model, a novel framework that simultaneously achieves clustering, subspace estimation, and variable selection, providing a unified solution for high-dimensional heterogeneous data analysis. We develop a group Lasso penalized expectation-maximization (EM) algorithm and obtain its non-asymptotic convergence rate. Through extensive simulation studies, mixPFC demonstrates superior performance compared to existing methods across various settings. Applications to real world datasets further highlight its effectiveness and practical advantages.

Paperid:447

Authors:Yuzhong Hong · Hanshan Zhang · Junwei Bao · Hongfei Jiang · yang song

Title: Energy-Based Preference Model Offers Better Offline Alignment than the Bradley-Terry Preference Model

Abstract:
Since the debut of DPO, it has been shown that aligning a target LLM with human preferences via the KLconstrained RLHF loss is mathematically equivalent to a special kind of reward modeling task. Concretely, the task requires: 1) using the target LLM to parameterize the reward model, and 2) tuning the reward model so that it has a 1:1 linear relationship with the true reward. However, we identify a significant issue: the DPO loss might have multiple minimizers, of which only one satisfies the required linearity condition. The problem arises from a well-known issue of the underlying Bradley-Terry preference model: it does not always have a unique maximum likelihood estimator (MLE). Consequently, the minimizer of the RLHF loss might be unattainable because it is merely one among many minimizers of the DPO loss. As a better alternative, we propose an energy-based preference model (EBM) that always has a unique MLE, inherently satisfying the linearity requirement. To showcase the practical utility of replacing BTM with our EBM in the context of offline alignment, we adapt a simple yet scalable objective function from the recent literature on fitting EBM and name it as Energy Preference Alignment (EPA). Empirically, we demonstrate that EPA consistently delivers better performance on open benchmarks compared to DPO, thereby validating the theoretical superiority of our EBM.

Paperid:448

Authors:Suning Huang · Zheyu Zhang · Tianhai Liang · Yihan Xu · Zhehao Kou · Chenhao Lu · Guowei Xu · Zhengrong Xue · Huazhe Xu

Title: MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning

Abstract:
Visual deep reinforcement learning (RL) enables robots to acquire skills from visual input for unstructured tasks. However, current algorithms suffer from low sample efficiency, limiting their practical applicability. In this work, we present MENTOR, a method that improves both the architecture and optimization of RL agents. Specifically, MENTOR replaces the standard multilayer perceptron (MLP) with a mixture-of-experts (MoE) backbone and introduces a task-oriented perturbation mechanism. MENTOR outperforms state-of-the-art methods across three simulation benchmarks and achieves an average of 83% success rate on three challenging real-world robotic manipulation tasks, significantly surpassing the 32% success rate of the strongest existing model-free visual RL algorithm. These results underscore the importance of sample efficiency in advancing visual RL for real-world robotics. Experimental videos are available at https://mentor-vrl.github.io/.

Paperid:449

Authors:Louis Airale · Antonio Longa · Mattia Rigon · Andrea Passerini · Roberto Passerone

Title: Simple Path Structural Encoding for Graph Transformers

Abstract:
Graph transformers extend global selfattention to graph-structured data, achieving notable success in graph learning. Recently, random walk structural encoding (RWSE) has been found to further enhance their predictive power by encoding both structural and positional information into the edge representation. However, RWSE cannot always distinguish between edges that belong to different local graph patterns, which reduces its ability to capture the full structural complexity of graphs.This work introduces Simple Path Structural Encoding (SPSE), a novel method that utilizes simple path counts for edge encoding. We show theoretically and experimentally that SPSE overcomes the limitations of RWSE, providing a richer representation of graph structures, particularly in capturing local cyclic patterns. To make SPSE computationally tractable, we propose an efficient approximate algorithm for simple path counting.SPSE demonstrates significant performance improvements over RWSE on various benchmarks, including molecular and long-range graph datasets, achieving statistically significant gains in discriminative tasks. These results pose SPSE as a powerful edge encoding alternative for enhancing the expressivity of graph transformers.

Paperid:450

Authors:Yinfeng Chen · Jin Liu · Rui Qiu

Title: Deep principal support vector machines for nonlinear sufficient dimension reduction

Abstract:
Abstract:The normal vectors obtained from the support vector machine (SVM) method offer the potential to achieve sufficient dimension reduction in both classification and regression scenarios. Motivated by it, we in this paper introduce a unified framework for nonlinear sufficient dimension reduction based on classification ensemble. Kernel principal SVM, a nonlinear sufficient dimension reduction method using the reproducing kernel Hilbert space and SVM, can almost be regarded as a special case of this framework. In particular, the neural network function class is considered to replace the reproducing kernel Hilbert space to implement a more flexible deep nonlinear sufficient dimension reduction. Its advantages are particularly prominent in cases of severe data contamination, large samples and complex inputs. Theoretically, we demonstrate that the optimal solution of the population objective function is unbiased in the sense of the central $\sigma$field. A nonasymptotic upper bound for the estimation error is also provided. Through simulations and real data analysis, our proposed deep nonlinear sufficient dimension reduction method exhibits considerable competitiveness.

Paperid:451

Authors:Jonathan Geuter · Gregor Kornhardt · Ingimar Tomasson · Vaios Laschos

Title: Universal Neural Optimal Transport

Abstract:
Optimal Transport (OT) problems are a cornerstone of many applications, but solving them is computationally expensive. To address this problem, we propose UNOT (Universal Neural Optimal Transport), a novel framework capable of accurately predicting (entropic) OT distances and plans between discrete measures of variable resolution for a given cost function. UNOT builds on Fourier Neural Operators, a universal class of neural networks that map between function spaces and that are discretizationinvariant, which enables our network to process measures of varying sizes. The network is trained adversarially using a second, generating network and a self-supervised bootstrapping loss. We theoretically justify the use of FNOs, prove that our generator is universal, and that minimizing the bootstrapping loss provably minimizes the ground truth loss. Through extensive experiments, we show that our network not only accurately predicts optimal transport distances and plans across a wide range of datasets, but also captures the geometry of the Wasserstein space correctly. Furthermore, we show that our network can be used as a state-of-the-art initialization for the Sinkhorn algorithm, significantly outperforming existing approaches.

Paperid:452

Authors:Emanuele Palumbo · Moritz Vandenhirtz · Alain Ryser · Imant Daunhawer · Julia Vogt

Title: From Logits to Hierarchies: Hierarchical Clustering made Simple

Abstract:
The hierarchical structure inherent in many realworld datasets makes the modeling of such hierarchies a crucial objective in both unsupervised and supervised machine learning. While recent advancements have introduced deep architectures specifically designed for hierarchical clustering, we adopt a critical perspective on this line of research. Our findings reveal that these methods face significant limitations in scalability and performance when applied to realistic datasets. Given these findings, we present an alternative approach and introduce a lightweight method that builds on pre-trained non-hierarchical clustering models. Remarkably, our approach outperforms specialized deep models for hierarchical clustering, and it is broadly applicable to any pre-trained clustering model that outputs logits, without requiring any fine-tuning. To highlight the generality of our approach, we extend its application to a supervised setting, demonstrating its ability to recover meaningful hierarchies from a pre-trained ImageNet classifier. Our results establish a practical and effective alternative to existing deep hierarchical clustering methods, with significant advantages in efficiency, scalability and performance.

Paperid:453

Authors:Eric Elmoznino · Tom Marty · Tejas Kasetty · Léo Gagnon · Sarthak Mittal · Mahan Fathi · Dhanya Sridhar · Guillaume Lajoie

Title: In-Context Learning and Occam's Razor

Abstract:
A central goal of machine learning is generalization. While the No Free Lunch Theorem states that we cannot obtain theoretical guarantees for generalization without further assumptions, in practice we observe that simple models which explain the training data generalize best—a principle called Occam's razor. Despite the need for simple models, most current approaches in machine learning only minimize the training error, and at best indirectly promote simplicity through regularization or architecture design. Here, we draw a connection between Occam's razor and incontext learning—an emergent ability of certain sequence models like Transformers to learn at inference time from past observations in a sequence. In particular, we show that the next-token prediction loss used to train in-context learners is directly equivalent to a data compression technique called prequential coding, and that minimizing this loss amounts to jointly minimizing both the training error and the complexity of the model that was implicitly learned from context. Our theory and the empirical experiments we use to support it not only provide a normative account of in-context learning, but also elucidate the shortcomings of current in-context learning methods, suggesting ways in which they can be improved.

Paperid:454

Authors:Aaron Wenteler · Martina Occhetta · Nikhil Branson · Victor Curean · Magdalena Huebner · William Dee · William Connell · Siu Chung · Alex Hawkins-Hooker · Yasha Ektefaie · César Miguel Valdez Córdova · Amaya Gallagher-Syed

Title: PertEval-scFM: Benchmarking Single-Cell Foundation Models for Perturbation Effect Prediction

Abstract:
In silicomodeling of transcriptional responses to perturbations is crucial for advancing our understanding of cellular processes and disease mechanisms. We present PertEvalscFM, a standardized framework designed to evaluate models for perturbation effect prediction. We apply PertEval-scFM to benchmark zero-shot single-cell foundation model (scFM) embeddings against simpler baseline models to assess whether these contextualized representations enhance perturbation effect prediction. Our results show that scFM embeddings do not provide consistent improvements over baseline models, especially under distribution shift. Additionally, all models struggle with predicting strong or atypical perturbation effects. Overall, this study provides a systematic evaluation of zero-shot scFM embeddings for perturbation effect prediction, highlighting the challenges of this task and revealing the limitations of current-generation scFMs. Our findings underscore the need for specialized models and high-quality datasets that capture a broader range of cellular states. Source code and documentation can be found at: https://anonymous.4open.science/r/PertEval-C674/README.md.

Paperid:455

Authors:Suyuan Liu · Hao Yu · Hao Tan · KE LIANG · Siwei Wang · Shengju Yu · En Zhu · Xinwang Liu

Title: Efficient Federated Incomplete Multi-View Clustering

Abstract:
Multiview clustering (MVC) leverages complementary information from diverse data sources to enhance clustering performance. However, its practical deployment in distributed and privacy-sensitive scenarios remains challenging. Federated multi-view clustering (FMVC) has emerged as a potential solution, but existing approaches suffer from substantial limitations, including excessive communication overhead, insufficient privacy protection, and inadequate handling of missing views. To address these issues, we propose Efficient Federated Incomplete Multi-View Clustering (EFIMVC), a novel framework that introduces a localized optimization strategy to significantly reduce communication costs while ensuring theoretical convergence. EFIMVC employs both view-specific and shared anchor graphs as communication variables, thereby enhancing privacy by avoiding the transmission of sensitive embeddings. To further improve robustness, we design a dual-anchor alignment mechanism that mitigates anchor misalignment during graph fusion. Moreover, EFIMVC seamlessly extends to scenarios with missing views, making it a practical and scalable solution for real-world applications. Extensive experiments on benchmark datasets demonstrate the superiority of EFIMVC in clustering accuracy, communication efficiency, and privacy preservation.

Paperid:456

Authors:Tianqi Du · Haotian Huang · Yifei Wang · Yisen Wang

Title: Output Alignment: A Fresh Perspective on Length Generalization in LLMs

Abstract:
Large language models (LLMs) have exhibited impressive performance and surprising emergent properties. However, their effectiveness is still limited by the fixed context window of the transformer architecture, posing challenges for length generalization. In this work, we propose a new perspective on length generalization by focusing on the output distribution, introducing a shift from the conventional emphasis on input features such as positional encodings or data structures. Specifically, through case studies on synthetic tasks, we highlight the critical role of \textbf{output alignment}the consistency of output distributions across sequences of varying lengths. Extending this insight to natural language tasks, we propose a metric called Long-Short Misalignment to quantify output alignment, uncovering a strong correlation between this metric and length generalization performance. Building on these findings, we develop a regularization loss that enhances output alignment during training. Extensive experiments validate the effectiveness of our approach, offering a fresh perspective for understanding and improving length generalization in LLMs.

Paperid:457

Authors:Guangting Zheng · Yehao Li · Yingwei Pan · Jiajun Deng · Ting Yao · Yanyong Zhang · Tao Mei

Title: Hierarchical Masked Autoregressive Models with Low-Resolution Token Pivots

Abstract:
Autoregressive models have emerged as a powerful generative paradigm for visual generation. The current defacto standard of next token prediction commonly operates over a single-scale sequence of dense image tokens, and is incapable of utilizing global context especially for early tokens prediction. In this paper, we introduce a new autoregressive design to model a hierarchy from a few low-resolution image tokens to the typical dense image tokens, and delve into a thorough hierarchical dependency across multi-scale image tokens. Technically, we present a Hierarchical Masked Autoregressive models (Hi-MAR) that pivot on low-resolution image tokens to trigger hierarchical autoregressive modeling in a multi-phase manner. Hi-MAR learns to predict a few image tokens in low resolution, functioning as intermediary pivots to reflect global structure, in the first phase. Such pivots act as the additional guidance to strengthen the next autoregressive modeling phase by shaping global structural awareness of typical dense image tokens. A new Diffusion Transformer head is further devised to amplify the global context among all tokens for mask token prediction. Extensive evaluations on both class-conditional and text-to-image generation tasks demonstrate that Hi-MAR outperforms typical AR baselines, while requiring fewer computational costs.

Paperid:458

Authors:Tianze Yang · Yucheng Shi · Mengnan Du · Xuansheng Wu · Qiaoyu Tan · Jin Sun · Ninghao Liu

Title: Concept-Centric Token Interpretation for Vector-Quantized Generative Models

Abstract:
VectorQuantized Generative Models (VQGMs) have emerged as powerful tools for image generation. However, the key component of VQGMs---the codebook of discrete tokens---is still not well understood, e.g., which tokens are critical to generate an image of a certain concept?This paper introduces Concept-Oriented Token Explanation (CORTEX), a novel approach for interpreting VQGMs by identifying concept-specific token combinations. Our framework employs two methods: (1) a sample-level explanation method that analyzes token importance scores in individual images, and (2) a codebook-level explanation method that explores the entire codebook to find globally relevant tokens. Experimental results demonstrate CORTEX's efficacy in providing clear explanations of token usage in the generative process, outperforming baselines across multiple pretrained VQGMs. Besides enhancing VQGMs transparency, CORTEX is useful in applications such as targeted image editing and shortcut feature detection.

Paperid:459

Authors:Markelle Kelly · Alex Boyd · Samuel Showalter · Padhraic Smyth · Mark Steyvers

Title: Bayesian Consensus Prediction for Correlated Human Experts and Classifiers

Abstract:
Abstract:Applications of machine learning often involve making predictions based on both model outputs and the opinions of human experts. In this context, we investigate the prediction of human expert consensus for $K$ary classification problems. Given a budget for querying experts, the objective is to achieve the maximum accuracy (in terms of predicting expert consensus) while leveraging the class probability estimates of pre-trained classifiers. We develop a general Bayesian framework for this problem, modeling expert correlation via a joint latent representation, enabling simulation-based inference about the utility of additional expert queries, as well as inference of posterior distributions over unobserved expert labels. We apply our approach to two real-world medical classification problems, as well as to CIFAR-10H and ImageNet-16H, demonstrating substantial reductions in the cost of querying human experts while maintaining high prediction accuracy.

Paperid:460

Authors:Rui Zhang · Yun Shen · Hongwei Li · Wenbo Jiang · Hanxiao Chen · Yuan Zhang · Guowen Xu · Yang Zhang

Title: The Ripple Effect: On Unforeseen Complications of Backdoor Attacks

Abstract:
Recent research highlights concerns about the trustworthiness of thirdparty Pre-Trained Language Models (PTLMs) due to potential backdoor attacks.These backdoored PTLMs, however, are effective only for specific pre-defined downstream tasks.In reality, these PTLMs can be adapted to many other unrelated downstream tasks.Such adaptation may lead to unforeseen consequences in downstream model outputs, consequently raising user suspicion and compromising attack stealthiness.We refer to this phenomenon as backdoor complications.In this paper, we undertake the first comprehensive quantification of backdoor complications.Through extensive experiments using 4 prominent PTLMs and 16 text classification benchmark datasets, we demonstrate the widespread presence of backdoor complications in downstream models fine-tuned from backdoored PTLMs.The output distribution of triggered samples significantly deviates from that of clean samples.Consequently, we propose a backdoor complication reduction method leveraging multi-task learning to mitigate complications without prior knowledge of downstream tasks.The experimental results demonstrate that our proposed method can effectively reduce complications while maintaining the efficacy and consistency of backdoor attacks.

Paperid:461

Authors:Jingbang Chen · Xinyuan Cao · Alicia Stepin · Li Chen

Title: On the Power of Learning-Augmented Search Trees

Abstract:
Abstract:We study learningaugmented binary search trees (BSTs) via Treaps with carefully designed priorities.The result is a simple search tree in which the depth of each item $x$ is determined by its predicted weight $w_x$.Specifically, each item $x$ is assigned a composite priority of $-\lfloor\log\log(1/w_x)\rfloor + U(0, 1)$ where $U(0, 1)$ is the uniform random variable. By choosing $w_x$ as the relative frequency of $x$, the resulting search trees achieve static optimality.This approach generalizes the recent learning-augmented BSTs [Lin-Luo-Woodruff ICML`22], which only work for Zipfian distributions, by extending them to arbitrary input distributions.Furthermore, we demonstrate that our method can be generalized to a B-Tree data structure using the B-Treap approach [Golovin ICALP'09]. Our search trees are also capable of leveraging localities in the access sequence through online self-reorganization, thereby achieving the working-set property. Additionally, they are robust to prediction errors and support dynamic operations, such as insertions, deletions, and prediction updates. We complement our analysis with an empirical study, demonstrating that our method outperforms prior work and classic data structures.

Paperid:462

Authors:Xubin Wang · Jianfei Wu · Yuan Yichen · Mingzhe Li · Deyu Cai · Weijia Jia

Title: Demonstration Selection for In-Context Learning via Reinforcement Learning

Abstract:
Diversity in demonstration selection is crucial for enhancing model generalization, as it enables a broader coverage of structures and concepts. However, constructing an appropriate set of demonstrations has remained a focal point of research. This paper presents the RelevanceDiversity Enhanced Selection (RDES), an innovative approach that leverages reinforcement learning to optimize the selection of diverse reference demonstrations for text classification tasks using Large Language Models (LLMs), especially in few-shot prompting scenarios. RDES employs a Q-learning framework to dynamically identify demonstrations that maximize both diversity and relevance to the classification objective by calculating a diversity score based on label distribution among selected demonstrations. This method ensures a balanced representation of reference data, leading to improved classification accuracy. Through extensive experiments on four benchmark datasets and involving 12 closed-source and open-source LLMs, we demonstrate that RDES significantly enhances classification accuracy compared to ten established baselines. Furthermore, we investigate the incorporation of Chain-of-Thought (CoT) reasoning in the reasoning process, which further enhances the model's predictive performance. The results underscore the potential of reinforcement learning to facilitate adaptive demonstration selection and deepen the understanding of classification challenges.

Paperid:463

Authors:David Boetius · Stefan Leue · Tobias Sutter

Title: Probabilistic Verification of Neural Networks using Branch and Bound

Abstract:
Probabilistic verification of neural networks is concerned with formally analysing the output distribution of a neural network under a probability distribution of the inputs. Examples of probabilistic verification include verifying the demographic parity fairness notion or quantifying the safety of a neural network. We present a new algorithm for the probabilistic verification of neural networks based on an algorithm for computing and iteratively refining lower and upper bounds on probabilities over the outputs of a neural network. By applying stateof-the-art bound propagation and branch and bound techniques from non-probabilistic neural network verification, our algorithm significantly outpaces existing probabilistic verification algorithms, reducing solving times for various benchmarks from the literature from tens of minutes to tens of seconds. Furthermore, our algorithm compares favourably even to dedicated algorithms for restricted subsets of probabilistic verification. We complement our empirical evaluation with a theoretical analysis, proving that our algorithm is sound and, under mildly restrictive conditions, also complete when using a suitable set of heuristics.

Paperid:464

Authors:Steinar Laenen · Peter Macgregor · He Sun

Title: Dynamic Similarity Graph Construction with Kernel Density Estimation

Abstract:
Abstract:In the kernel density estimation (KDE) problem, we are given a set $X$ of data points in $\mathbb{R}^d$, a kernel function $k: \mathbb{R}^d \times \mathbb{R}^d \rightarrow \mathbb{R}$, and a query point $\mathbf{q} \in \mathbb{R}^d$, and the objective is to quickly output an estimate of $\sum_{\mathbf{x} \in X} k(\mathbf{q}, \mathbf{x})$.In this paper, we consider $\textsf{KDE}$ in the dynamic setting, and introduce a data structure that efficiently maintains the _estimates_ for a set of query points as data points are added to $X$ over time.Based on this, we design a dynamic data structure that maintains a sparse approximation of the fully connected similarity graph on $X$, and develop a fast dynamic spectral clustering algorithm.We further evaluate the effectiveness of our algorithms on both synthetic and realworld datasets.

Paperid:465

Authors:Swetha Ganesh · Washim Mondal · Vaneet Aggarwal

Title: A Sharper Global Convergence Analysis for Average Reward Reinforcement Learning via an Actor-Critic Approach

Abstract:
Abstract:This work examines averagereward reinforcement learning with general policy parametrization. Existing state-of-the-art (SOTA) guarantees for this problem are either suboptimal or hindered by several challenges, including poor scalability with respect to the size of the state-action space, high iteration complexity, and dependence on knowledge of mixing times and hitting times. To address these limitations, we propose a Multi-level Monte Carlo-based Natural Actor-Critic (MLMC-NAC) algorithm. Our work is the first to achieve a global convergence rate of $\Tilde{\mathcal{O}}(1/\sqrt{T})$ for average-reward Markov Decision Processes (MDPs) (where $T$ is the horizon length), without requiring the knowledge of mixing and hitting times. Moreover, the convergence rate does not scale with the size of the state space, therefore even being applicable to infinite state spaces.

Paperid:466

Authors:Boyuan Wu · wang · Xianwei Lin · Jiachun Xu · Jikai Yu · Zhou Shicheng · Hongda Chen · Lianxin Hu

Title: Distributed Parallel Gradient Stacking(DPGS): Solving Whole Slide Image Stacking Challenge in Multi-Instance Learning

Abstract:
Whole Slide Image (WSI) analysis is framed as a Multiple Instance Learning (MIL) problem, but existing methods struggle with nonstackable data due to inconsistent instance lengths, which degrades performance and efficiency. We propose a Distributed Parallel Gradient Stacking (DPGS) framework with Deep Model-Gradient Compression (DMGC) to address this. DPGS enables lossless MIL data stacking for the first time, while DMGC accelerates distributed training via joint gradient-model compression. Experiments on Camelyon16 and TCGA-Lung datasets demonstrate up to 31× faster training, up to a 99.2% reduction in model communication size at convergence, and up to a 9.3% improvement in accuracy compared to the baseline. To our knowledge, this is the first work to solve non-stackable data in MIL while improving both speed and accuracy.

Paperid:467

Authors:Ankit Ghosh · Gargee Kashyap · Sarthak Mittal · Nupur Jain · Raghavan B Sunoj · Abir De

Title: Learning Condensed Graph via Differentiable Atom Mapping for Reaction Yield Prediction

Abstract:
Yield of chemical reactions generally depends on the activation barrier, i.e., the energy difference between the reactant and the transition state. Computing the transition state from the reactant and product graphs requires prior knowledge of the correct node alignment (i.e., atom mapping), which is not available in yield prediction datasets. In this work, we propose YieldNet, a neural yield prediction model, which tackles these challenges. Here, we first approximate the atom mapping between the reactants and products using a differentiable node alignment network. We then use this approximate atom mapping to obtain a noisy realization of the condensed graph of reaction (CGR), which is a supergraph encompassing both the reactants and products. This CGR serves as a surrogate for the transition state graph structure. The CGR embeddings of different steps in a multistep reaction are then passed into a transformer-guided reaction path encoder.Our experiments show that YieldNet can predict the yield more accurately than the baselines. Furthermore, the model is trained only under the distant supervision of yield values, without requiring fine-grained supervision of atom mapping.

Paperid:468

Authors:Filip Szatkowski · Yaoyue Zheng · Fei Yang · Tomasz Trzcinski · Bartłomiej Twardowski · Joost van de Weijer

Title: Improving Continual Learning Performance and Efficiency with Auxiliary Classifiers

Abstract:
Continual learning is crucial for applying machine learning in challenging, dynamic, and often resourceconstrained environments. However, catastrophic forgetting — overwriting previously learned knowledge when new information is acquired — remains a major challenge. In this work, we examine the intermediate representations in neural network layers during continual learning and find that such representations are less prone to forgetting, highlighting their potential to accelerate computation. Motivated by these findings, we propose to use auxiliary classifiers~(ACs) to enhance performance and demonstrate that integrating ACs into various continual learning methods consistently improves accuracy across diverse evaluation settings, yielding an average 10\% relative gain. We also leverage the ACs to reduce the average cost of the inference by 10-60\% without compromising accuracy, enabling the model to return the predictions before computing all the layers. Our approach provides a scalable and efficient solution for continual learning.

Paperid:469

Authors:Jerry Huang · Peng Lu · QIUHAO Zeng

Title: Calibrated Language Models and How to Find Them with Label Smoothing

Abstract:
Recent advances in natural language processing (NLP) have opened up greater opportunities to enable finetuned large language models (LLMs) to behave as more powerful interactive agents through improved instruction-following ability. However, understanding how this impacts confidence calibration for reliable model output has not been researched in full. In this work, we examine various open-sourced LLMs, identifying significant calibration degradation after instruction tuning in each. Seeking a practical solution, we look towards label smoothing, which has been shown as an effective method to regularize for overconfident predictions but has yet to be widely adopted in the supervised fine-tuning (SFT) of LLMs. We first provide insight as to why label smoothing is sufficient to maintain calibration throughout the SFT process. However, settings remain where the effectiveness of smoothing is severely diminished, in particular the case of large vocabulary LLMs (LV-LLMs). We posit the cause to stem from the ability to become over-confident, which has a direct relationship with the hidden size and vocabulary size, and justify this theoretically and experimentally. Finally, we address an outstanding issue regarding the memory footprint of the cross-entropy loss computation in the label smoothed loss setting, designing a customized kernel to dramatically reduce memory consumption without sacrificing speed or performance in comparison to existing solutions for non-smoothed losses.

Paperid:470

Authors:Francesco Emanuele Stradi · Anna Lunghi · Matteo Castiglioni · Alberto Marchesi · Nicola Gatti

Title: Policy Optimization for CMDPs with Bandit Feedback: Learning Stochastic and Adversarial Constraints

Abstract:
Abstract:We study online learning in constrained Markov decision processes (CMDPs) in which rewards and constraints may be either stochastic or adversarial. In such settings, stradi et al. (2024) proposed the first bestof-both-worlds algorithm able to seamlessly handle stochastic and adversarial constraints, achieving optimal regret and constraint violation bounds in both cases. This algorithm suffers from two major drawbacks. First, it only works under full feedback, which severely limits its applicability in practice. Moreover, it relies on optimizing over the space of occupancy measures, which requires solving convex optimization problems, an highly inefficient task. In this paper, we provide the first best-of-both-worlds algorithm for CMDPs with bandit feedback. Specifically, when the constraints are stochastic, the algorithm achieves $\widetilde{\mathcal{O}}(\sqrt{T})$ regret and constraint violation, while, when they are adversarial, it attains $\widetilde{\mathcal{O}}(\sqrt{T})$ constraint violation and a tight fraction of the optimal reward. Moreover, our algorithm is based on a policy optimization approach, which is much more efficient than occupancy-measure-based methods.

Paperid:471

Authors:Zhiqiang Wang · Jianghao Wen · Jianqing Liang

Title: Delay-DSGN: A Dynamic Spiking Graph Neural Network with Delay Mechanisms for Evolving Graph

Abstract:
Dynamic graph representation learning using Spiking Neural Networks (SNNs) exploits the temporal spiking behavior of neurons, offering advantages in capturing the temporal evolution and sparsity of dynamic graphs. However, existing SNNbased methods often fail to effectively capture the impact of latency in information propagation on node representations. To address this, we propose Delay-DSGN, a dynamic spiking graph neural network incorporating a learnable delay mechanism. By leveraging synaptic plasticity, the model dynamically adjusts connection weights and propagation speeds, enhancing temporal correlations and enabling historical data to influence future representations. Specifically, we introduce a Gaussian delay kernel into the neighborhood aggregation process at each time step, adaptively delaying historical information to future time steps and mitigating information forgetting. Experiments on three large-scale dynamic graph datasets demonstrate that Delay-DSGN outperforms eight state-of-the-art methods, achieving the best results in node classification tasks. We also theoretically derive the constraint conditions between the Gaussian kernel's standard deviation and size, ensuring stable training and preventing gradient explosion and vanishing issues.

Paperid:472

Authors:Robert Busa-Fekete · Travis Dick · Claudio Gentile · Haim Kaplan · Tomer Koren · Uri Stemmer

Title: Nearly Optimal Sample Complexity for Learning with Label Proportions

Abstract:
We investigate Learning from Label Proportions (LLP), a partial information setting where examples in a training set are grouped into bags, and only aggregate label values in each bag are available. Despite the partial observability, the goal is still to achieve small regret at the level of individual examples. We give results on the sample complexity of LLP under square loss, showing that our sample complexity is essentially optimal. From an algorithmic viewpoint, we rely on carefully designed variants of Empirical Risk Minimization, and Stochastic Gradient Descent algorithms, combined with ad hoc variance reduction techniques. On one hand, our theoretical results improve in important ways on the existing literature on LLP, specifically in the way the sample complexity depends on the bag size. On the other hand, we validate our algorithmic solutions on several datasets, demonstrating improved empirical performance (better accuracy for less samples) against recent baselines.

Paperid:473

Authors:Yusuke Yamasaki · Kenta Niwa · Takumi Fukami · Takayuki Miura · Daiki Chijiwa

Title: Plausible Token Amplification for Improving Accuracy of Differentially Private InContext Learning Based on Implicit Bayesian Inference

Abstract:
We propose Plausible Token Amplification (PTA) to improve the accuracy of Differentially Private InContext Learning (DP-ICL) using DP synthetic demonstrations. While Tang et al. empirically improved the accuracy of DP-ICL by limiting vocabulary space during DP synthetic demonstration generation, its theoretical basis remains unexplored. By interpreting ICL as implicit Bayesian inference on a concept underlying demonstrations, we not only provide theoretical evidence supporting Tang et al.'s empirical method but also introduce PTA, a refined method for modifying next-token probability distribution. Through the modification, PTA highlights tokens that distinctly represent the ground-truth concept underlying the original demonstrations. As a result, generated DP synthetic demonstrations guide the Large Language Model to successfully infer the ground-truth concept, which improves the accuracy of DP-ICL. Experimental evaluations on both synthetic and real-world text-classification datasets validated the effectiveness of PTA.

Paperid:474

Authors:Chong Zhang · Peng Zhang · Mengke Li

Title: Angle-Robust Networks based on Gamma Distribution PCA for SAR Target Recognition

Abstract:
Abstract:Scattering characteristics of synthetic aperture radar (SAR) targets are typically related to observed azimuth and depression angles. However, in practice, it is difficult to obtain adequate training samples at all observation angles, which probably leads to poor robustness of deep networks. In this paper, we first propose a GammaDistribution Principal Component Analysis ($\Gamma$PCA) model that fully accounts for the statistical characteristics of SAR data. The $\Gamma$PCA derives consistency convolution kernels to effectively capture the angle-invariant features of the same target at various attitude angles, thus alleviating deep models' sensitivity to angle changes in SAR target recognition task. We validate $\Gamma$PCA model based on two different backbones, ResNet and ViT, and conducted multiple robustness experiments on the MSTAR benchmark dataset. The experimental results demonstrated that $\Gamma$PCA effectively enables the model to withstand substantial distributional discrepancy caused by angle changes. Additionally, $\Gamma$PCA convolution kernel is designed to require no parameter updates, thereby bringing no computational burden to a network.

Paperid:475

Authors:Ziyang Wu · Jingyuan Zhang · Druv Pai · XuDong Wang · Chandan Singh · Jianwei Yang · Jianfeng Gao · Yi Ma

Title: Simplifying DINO by Coding Rate Regularization

Abstract:
DINO and DINOv2 are two model families being widely used to learn representations from unlabeled imagery data at large scales. Their learned representations often enable stateof-the-art performance for downstream tasks, such as image classification and segmentation. However, they employ many empirically motivated design choices and their training pipelines are highly complex and unstable --- many hyperparameters need to be carefully tuned to ensure that the representations do not collapse --- which poses considerable difficulty to improving them or adapting them to new domains. In this work, we posit that we can remove most such-motivated idiosyncrasies in the pre-training pipelines, and only need to add an explicit coding rate term in the loss function to avoid collapse of the representations. As a result, we obtain highly simplified variants of the DINO and DINOv2 which we call SimDINO and SimDINOv2, respectively. Remarkably, these simplified models are more robust to different design choices, such as network architecture and hyperparameters, and they learn even higher-quality representations, measured by performance on downstream tasks, offering a Pareto improvement over the corresponding DINO and DINOv2 models. This work highlights the potential of using simplifying design principles to improve the empirical practice of deep learning.

Paperid:476

Authors:Chenyou Fan · Chenjia Bai · Zhao Shan · Haoran He · Yang Zhang · Zhen Wang

Title: Task-agnostic Pre-training and Task-guided Fine-tuning for Versatile Diffusion Planner

Abstract:
Diffusion models have demonstrated their capabilities in modeling trajectories of multitasks. However, existing multi-task planners or policies typically rely on task-specific demonstrations via multi-task imitation, or require task-specific reward labels to facilitate policy optimization via Reinforcement Learning (RL). They are costly due to the substantial human efforts required to collect expert data or design reward functions. To address these challenges, we aim to develop a versatile diffusion planner capable of leveraging large-scale inferior data that contains task-agnostic sub-optimal trajectories, with the ability to fast adapt to specific tasks. In this paper, we propose SODP, a two-stage framework that leverages Sub-Optimal data to learn a Diffusion Planner, which is generalizable for various downstream tasks. Specifically, in the pre-training stage, we train a foundation diffusion planner that extracts general planning capabilities by modeling the versatile distribution of multi-task trajectories, which can be sub-optimal and has wide data coverage. Then for downstream tasks, we adopt RL-based fine-tuning with task-specific rewards to quickly refine the diffusion planner, which aims to generate action sequences with higher task-specific returns. Experimental results from multi-task domains including Meta-World and Adroit demonstrate that SODP outperforms state-of-the-art methods with only a small amount of data for reward-guided fine-tuning.

Paperid:477

Authors:Yuxuan Zhou · Xien Liu · Chenwei Yan · Chen Ning · Xiao Zhang · Boxun Li · Xiangling Fu · Shijin Wang · Guoping Hu · Yu Wang · Ji Wu

Title: Evaluating LLMs Across Multi-Cognitive Levels: From Medical Knowledge Mastery to Scenario-Based Problem Solving

Abstract:
Large language models (LLMs) have demonstrated remarkable performance on various medical benchmarks, but their capabilities across different cognitive levels remain underexplored. Inspired by Bloom's Taxonomy, we propose a multicognitive-level evaluation framework for assessing LLMs in the medical domain in this study. The framework integrates existing medical datasets and introduces tasks targeting three cognitive levels: preliminary knowledge grasp, comprehensive knowledge application, and scenario-based problem solving. Using this framework, we systematically evaluate state-of-the-art general and medical LLMs from five prominent families: Llama, Qwen, Gemma, Phi, and GPT. Our findings reveal a significant performance decline as cognitive complexity increases across evaluated models, with model size playing a more critical role in performance at higher cognitive levels. Our study highlights the need to enhance LLMs' medical capabilities at higher cognitive levels and provides insights for developing LLMs suited to real-world medical applications.

Paperid:478

Authors:Mingjia Li · Hong Qian · Tian-Zuo Wang · ShujunLi · Min Zhang · Aimin Zhou

Title: Strong and Weak Identifiability of Optimization-based Causal Discovery

Abstract:
Causal discovery aims to identify causal relationships from observational data. It is crucial in various scientific fields as it helps to understand underlying mechanisms. Recently, optimizationbased causal discovery methods have attracted extensive attention in the literature due to their efficiency in handling high-dimensional problems. However, we observe that optimization-based methods often perform well on certain problems but struggle with others. This paper identifies a specific characteristic of causal structural equations that determines the difficulty of identification in causal discovery and, in turn, the performance of optimization-based methods. We conduct an in-depth study of the additive noise model (ANM) and propose to further divide identifiable problems into strongly and weakly identifiable types based on the difficulty of identification. We also provide a sufficient condition to distinguish between these two categories. Inspired by these findings, this paper further proposes GENE, a generic method to address both strongly and weakly identifiable problems in a unified way under the ANM assumption. GENE adopts an order-based search framework that incorporates conditional independence tests into the order fitness evaluation, ensuring effectiveness on weakly identifiable problems. Besides, the proposed order-based optimization method restricts the dimensionality of effect variables to ensure scale invariance, a property crucial for practical applications. Experiments on both synthetic and real-world datasets demonstrate that GENE is uniquely effective in addressing weakly identifiable problems. For strongly identifiable problems, GENE remains competitive with state-of-the-art causal discovery algorithms.

Paperid:479

Authors:Zhenglin Wan · Xingrui Yu · David Bossens · Yueming LYU · Qing Guo · Flint Xiaofeng Fan · Yew Soon ONG · Ivor Tsang

Title: Diversifying Robot Locomotion Behaviors with Extrinsic Behavioral Curiosity

Abstract:
Imitation learning (IL) has shown promise in robot locomotion but is often limited to learning a single expert policy, constraining behavior diversity and robustness in unpredictable realworld scenarios. To address this, we introduce Quality Diversity Inverse Reinforcement Learning (QD-IRL), a novel framework that integrates quality-diversity optimization with IRL methods, enabling agents to learn diverse behaviors from limited demonstrations. This work introduces Extrinsic Behavioral Curiosity (EBC), which allows agents to receive additional curiosity rewards from an external critic based on how novel the behaviors are with respect to a large behavioral archive. To validate the effectiveness of EBC in exploring diverse locomotion behaviors, we evaluate our method on multiple robot locomotion tasks. EBC improves the performance of QD-IRL instances with GAIL, VAIL, and DiffAIL by up to 185\%, 42\%, and 150\% in Humanoid environments, even surpassing expert performance by 20\%. Furthermore, we demonstrate that EBC is applicable to Quality Diversity Reinforcement Learning (QD-RL), where it substantially improves performance and provides a generic technique for diverse robot locomotion.

Paperid:480

Authors:Iulia Duta · Pietro Lió

Title: SPHINX: Structural Prediction using Hypergraph Inference Network

Abstract:
The importance of higherorder relations is widely recognized in numerous real-world systems. However, annotating them is a tedious and sometimes even impossible task. Consequently, current approaches for data modelling either ignore the higher-order interactions altogether or simplify them into pairwise connections. To facilitate higher-order processing, even when a hypergraph structure is not available, we introduce SPHINX, a model that learns to infer a latent hypergraph structure in an unsupervised way, solely from the final task-dependent signal. To ensure broad applicability, we design the model to be end-to-end differentiable, capable of generating a discrete hypergraph structure compatible with any modern hypergraph networks, and easily optimizable without requiring additional regularization losses.Through extensive ablation studies and experiments conducted on four challenging datasets, we demonstrate that our model is capable of inferring suitable latent hypergraphs in both transductive and inductive tasks. Moreover, the inferred latent hypergraphs are interpretable and contribute to enhancing the final performance, outperforming existing methods for hypergraph prediction.

Paperid:481

Authors:Shi Yin · Xinyang Pan · fengyan wang · Lixin He

Title: TraceGrad: a Framework Learning Expressive SO(3)-equivariant Non-linear Representations for Electronic-Structure Hamiltonian Prediction

Abstract:
We propose a framework to combine strong nonlinear expressiveness with strict SO(3)-equivariance in prediction of the electronic-structure Hamiltonian, by exploring the mathematical relationships between SO(3)-invariant and SO(3)-equivariant quantities and their representations. The proposed framework, calledTraceGrad, first constructs theoretical SO(3)-invarianttracequantities derived from the Hamiltonian targets, and use these invariant quantities as supervisory labels to guide the learning of high-quality SO(3)-invariant features. Given that SO(3)-invariance is preserved under non-linear operations, the learning of invariant features can extensively utilize non-linear mappings, thereby fully capturing the non-linear patterns inherent in physical systems. Building on this, we propose agradient-based mechanism to induce SO(3)-equivariant encodings of various degrees from the learned SO(3)-invariant features. This mechanism can incorporate powerful non-linear expressive capabilities into SO(3)-equivariant features with consistency of physical dimensions to the regression targets, while theoretically preserving equivariant properties, establishing a strong foundation for predicting Hamiltonian. Our method achieves state-of-the-art performance in prediction accuracy across eight challenging benchmark databases on Hamiltonian prediction. Experimental results demonstrate that this approach not only improves the accuracy of Hamiltonian prediction but also significantly enhances the prediction for downstream physical quantities, and also markedly improves the acceleration performance for the traditional Density Functional Theory algorithms.

Paperid:482

Authors:Zhining Liu · Ze Yang · Xiao Lin · Ruizhong Qiu · Tianxin Wei · Yada Zhu · Hendrik Hamann · Jingrui He · Hanghang Tong

Title: Breaking Silos: Adaptive Model Fusion Unlocks Better Time Series Forecasting

Abstract:
Timeseries forecasting plays a critical role in many real-world applications. Although increasingly powerful models have been developed and achieved superior results on benchmark datasets, through a fine-grainedsample-level inspection, we find that (i) no single model consistently outperforms others across different test samples, but instead (ii) each model excels in specific cases. These findings prompt us to explore how to adaptively leverage the distinct strengths of various forecasting models for different samples. We introduce TimeFuse, a framework for collective time-series forecasting withsample-level adaptive fusionof heterogeneous models. TimeFuse utilizes meta-features to characterize input time series and trains a learnable fusor to predict optimal model fusion weights for any given input. The fusor can leverage samples from diverse datasets for joint training, allowing it to adapt to a wide variety of temporal patterns and thus generalize to new inputs, even from unseen datasets. Extensive experiments demonstrate the effectiveness of TimeFuse in various long-/short-term forecasting tasks, achieving near-universal improvement over the state-of-the-art individual models.

Paperid:483

Authors:Junyan Li · Yang Zhang · Muhammad Yusuf Hassan · Talha Chafekar · Tianle Cai · Zhile Ren · Pengsheng Guo · Foroozan Karimzadeh · Colorado Reed · Chong Wang · Chuang Gan

Title: CommVQ: Commutative Vector Quantization for KV Cache Compression

Abstract:
Large Language Models (LLMs) are increasingly used in applications requiring long context lengths, but the keyvalue (KV) cache often becomes a memory bottleneck on GPUs as context lengths grow. To address this, we proposeCommutativeVectorQuantization (CommVQ) to significantly reduce memory usage for long-context LLM inference. First, we leverage additive quantization by introducing a lightweight encoder and codebook to compress the KV cache, which can then be decoded with a simple matrix multiplication. Second, to tackle the high computational costs during decoding, we design the codebook to be commutative with Rotary Position Embedding (RoPE), and utilize an Expectation-Maximization (EM) algorithm to learn the codebook. This enables efficient integration of decoding into the self-attention mechanism, significantly reducing computational overhead. Our approach achieves superior accuracy through additive quantization while lowering computational costs with our RoPE-commutative codebook. Experiments on long-context benchmarks and GSM8K demonstrate that our method reduces FP16 KV cache size by 87.5% while maintaining higher accuracy than state-of-the-art KV cache quantization methods. Remarkably, it enables 1-bit quantization of the KV cache with minimal accuracy degradation, making it possible to run LLaMA-3.1 8B model with a 500K context length on a single 4090 GPU.

Paperid:484

Authors:Qian Peng · Yajie Bao · Haojie Ren · Zhaojun Wang · Changliang Zou

Title: Conformal Prediction with Cellwise Outliers: A Detect-then-Impute Approach

Abstract:
Abstract:Conformal prediction is a powerful tool for constructing prediction intervals for blackbox models, providing a finite sample coverage guarantee for exchangeable data. However, this exchangeability is compromised when some entries of the test feature are contaminated, such as in the case of cellwise outliers. To address this issue, this paper introduces a novel framework called *detect-then-impute conformal prediction*. This framework first employs an outlier detection procedure on the test feature and then utilizes an imputation method to fill in those cells identified as outliers. To quantify the uncertainty in the processed test feature, we adaptively apply the detection and imputation procedures to the calibration set, thereby constructing exchangeable features for the conformal prediction interval of the test label. We develop two practical algorithms, $\texttt{PDI-CP}$ and $\texttt{JDI-CP}$, and provide a distribution-free coverage analysis under some commonly used detection and imputation procedures. Notably, $\texttt{JDI-CP}$ achieves a finite sample $1-2\alpha$ coverage guarantee. Numerical experiments on both synthetic and real datasets demonstrate that our proposed algorithms exhibit robust coverage properties and comparable efficiency to the oracle baseline.

Paperid:485

Authors:Firas Laakom · Haobo Chen · Yuheng Bu · Jürgen Schmidhuber

Title: Fairness Overfitting in Machine Learning: An Information-Theoretic Perspective

Abstract:
Despite substantial progress in promoting fairness in highstake applications using machine learning models, existing methods often modify the training process—such as through regularizers or other interventions—but lack formal guarantees that fairness achieved during training will generalize to unseen data. Although overfitting with respect to prediction performance has been extensively studied, overfitting in terms of fairness loss has received far less attention. In this paper, we propose a theoretical framework for analyzing fairness generalization error through an information-theoretic lens. Our novel bounding technique is based on Efron–Stein inequality, which allows us to derive tight information-theoretic fairness generalization bounds with both Mutual Information (MI) and Conditional Mutual Information (CMI). Our empirical results validate the tightness and practical relevance of these bounds across diverse fairness-aware learning algorithms. Our framework offers valuable insights to guide the design of algorithms improving fairness generalization.

Paperid:486

Authors:Hongcheng Guo · Yue Wang · Shaosheng Cao · Fei zhao · Boyang Wang · Lei Li · Liang Chen · Xinze Lyu · Zhe Xu · Yao Hu · Zhoujun Li

Title: SNS-Bench: Defining, Building, and Assessing Capabilities of Large Language Models in Social Networking Services

Abstract:
With the rapid advancement of Social Networking Services (SNS), the need for intelligent and efficient interaction within diverse platforms has become more crucial. Large Language Models (LLMs) play an importent role in SNS as they possess the potential to revolutionize user experience, content generation, and communication dynamics. However, recent studies focus on isolated SNS tasks rather than a comprehensive evaluation.In this paper, we introduce SNSBench, specially constructed for assessing the abilities of large language models from different \textbf{S}ocial \textbf{N}etworking \textbf{S}ervices, with a wide range of SNS-related information. SNS-Bench encompasses 8 different tasks such as note classification, query content relevance, and highlight words generation in comments. Finally, 6,658 questions of social media text, including subjective questions, single-choice, and multiple-choice questions, are concluded in SNS-Bench. Further, we evaluate the performance of over 25+ current diverse LLMs on our SNS-Bench. Models with different sizes exhibit performance variations, yet adhere to the scaling law.Moreover, we hope provide more insights to revolutionize the techniques of social network services with LLMs

Paperid:487

Authors:Dooho Lee · Myeong Kong · Sagad Hamid · Cheonwoo Lee · Jaemin Yoo

Title: Aggregation Buffer: Revisiting DropEdge with a New Parameter Block

Abstract:
We revisit DropEdge, a data augmentation technique for GNNs which randomly removes edges to expose diverse graph structures during training. While being a promising approach to effectively reduce overfitting on specific connections in the graph, we observe that its potential performance gain in supervised learning tasks is significantly limited. To understand why, we provide a theoretical analysis showing that the limited performance of DropEdge comes from the fundamental limitation that exists in many GNN architectures.Based on this analysis, we proposeAggregation Buffer, a parameter block specifically designed to improve the robustness of GNNs by addressing the limitation of DropEdge. Our method is compatible with any GNN model, and shows consistent performance improvements on multiple datasets. Moreover, our method effectively addresses wellknown problems such as degree bias or structural disparity as a unifying solution. Code and datasets are available at https://anonymous.4open.science/r/AggregationBuffer-46B4.

Paperid:488

Authors:Guy Hacohen · Tinne Tuytelaars

Title: Predicting the Susceptibility of Examples to Catastrophic Forgetting

Abstract:
Catastrophic forgetting the tendency of neural networks to forget previously learned data when learning new information - remains a major challenge in continual learning. In this work, we take a behavioral perspective, observing a connection between learning speed and forgetting: examples learned more quickly tend to be more resistant to forgetting. We then study replay-based continual learning and show that the choice of which examples populate the replay buffer - specifically, whether they were learned quickly or slowly - significantly impacts forgetting. Building on these insights, we introduce Speed-Based Sampling (SBS), a simple yet general strategy that selects replay examples based on their learning speed. SBS integrates seamlessly into existing buffer-based methods and consistently improves performance across a wide range of competitive continual learning benchmarks, advancing state-of-the-art results. Our findings highlight the importance of considering example susceptibility to forgetting when designing continual learning algorithms.

Paperid:489

Authors:Yi Xie · Jaedong Hwang · Carlos Brody · David Tank · Ila R. Fiete

Title: A Multi-Region Brain Model to Elucidate the Role of Hippocampus in Spatially Embedded Decision-Making

Abstract:
Brains excel at robust decisionmaking and data-efficient learning. Understanding the architectures and dynamics underlying these capabilities can inform inductive biases for deep learning. We present a multi-region brain model that explores the normative role of structured memory circuits in a spatially embedded binary decision-making task.We counterfactually compare the learning performance and neural representations of reinforcement learning (RL) agents with brain models of different interaction architectures between grid and place cells in the entorhinal cortex and hippocampus, coupled with an action-selection cortical recurrent neural network. We demonstrate that a specific architecture--where grid cells receive and jointly encode abstract velocity signals for self-movement and decision evidence increments--optimizes learning efficiency while best reproducing experimental observations relative to alternative architectures.Our findings therefore suggest brain-inspired structured architectures for efficient RL. Importantly, the models make novel, testable predictions about the organization and information flow within the entorhinal-hippocampal-neocortical circuitry: we predict that grid cells must conjunctively encode position and evidence for effective spatial decision-making, directly motivating new neurophysiological experiments.

Paperid:490

Authors:Yuchen Lin · Ronan Le Bras · Kyle Richardson · Ashish Sabharwal · Radha Poovendran · Peter Clark · Yejin Choi

Title: ZebraLogic: On the Scaling Limits of LLMs for Logical Reasoning

Abstract:
We investigate the logical reasoning capabilities of Large Language Models (LLMs) and their scalability across complex deductive tasks. Using ZebraLogic, a newly developed benchmark dataset of logic grid puzzles derived from constraint satisfaction problems (CSPs), we systematically evaluate LLM performance. ZebraLogic spans a broad range of search space complexities and incorporates diverse logical constraints, providing a controlled environment to assess reasoning abilities. Our results reveal a significant decline in accuracy as problem complexity increases—a phenomenon we term the “curse of complexity.” Notably, this limitation persists even with scaling model size and inferencetime computation, suggesting fundamental constraints in current LLM reasoning capabilities. Additionally, we explore strategies such as Best-of-N sampling, backtracking mechanisms, and self-verification prompts to enhance logical reasoning performance. Our findings provide critical insights into the scaling behavior of LLMs, highlight their limitations, and outline potential directions for advancing their reasoning capabilities.

Paperid:491

Authors:Aaron Mueller · Atticus Geiger · Sarah Wiegreffe · Dana Arad · Iván Arcuschin · Adam Belfki · Yik Siu Chan · Jaden Fiotto-Kaufman · Tal Haklay · Michael Hanna · Jing Huang · Rohan Gupta · Yaniv Nikankin · Hadas Orgad · Nikhil Prakash · Anja Reusch · Aruna Sankaranarayanan · Shun Shao · Alessandro Stolfo · Martin Tutek · Amir Zur · David Bau · Yonatan Belinkov

Title: MIB: A Mechanistic Interpretability Benchmark

Abstract:
Progress in mechanistic interpretability has been rapid, but there exist few standards for comparing methods. How can we know whether new methods achieve real improvements? In pursuit of meaningful and lasting evaluation standards, we propose MIB, a Mechanistic Interpretability Benchmark with two tracks spanning four tasks and five models. MIB favors methods that precisely and concisely recover relevant causal pathways or specific causal variables in neural language models. The circuit localization track compares methods that locate the model components—and connections between them—most important for performing a task (e.g., attribution patching or information flow routes).The causal variable localization track compares methods that featurize a hidden vector (e.g., sparse autoencoders or distributed alignment search) and locate mediators of a causal variable relevant to the task. Using MIB, we find that attribution and mask optimization methods perform best on circuit localization. We also find that supervised methods for causal variable localization outperform unsupervised methods, and both outperform the simplest baseline. These findings illustrate that MIB enables meaningful comparisons of MI methods, and increases our confidence that there has been real progress in the field.

Paperid:492

Authors:Alireza Naderi · Thiziri Nait Saada · Jared Tanner

Title: Mind the Gap: a Spectral Analysis of Rank Collapse and Signal Propagation in Attention Layers

Abstract:
Attention layers are the core component of transformers, the current stateof-the-art neural network architecture. Alternatives to softmax-based attention are being explored due to its tendency to hinder effective information flow. Evenat initialisation, it remains poorly understood why the propagation of signals and gradients through these random networks can be pathological, resulting in issues known as (i) vanishing/exploding gradients and (ii) rank collapsein depth, i.e. when all tokens converge to a single representation along layers. While rank collapse in depth naturally arises from repeated matrix multiplications---a common pattern across various architectures---we identify an additional and previously unknown challenge unique to softmax attention layers: (iii) rank collapsein width, which occurs as the context length increases. Using Random Matrix Theory, we conduct a rigorous analysis that uncovers a spectral gap between the two largest singular values of the attention matrix as the cause of (iii), which in turn exacerbates (i) and (ii).Building on this insight, we propose a novel yet simple practical solution to mitigate rank collapse in width by removing the outlier eigenvalue(s). Our theoretical framework offers a fresh perspective on recent practical studies, such as (Ye et al., 2024; Ali et al., 2023), whose ad hoc solutions can now be interpreted as implicit efforts to address the spectral gap issue. This work provides valuable theoretical support for ongoing large-scale empirical research, bringing theory and practice one step closer in the understanding of transformers.

Paperid:493

Authors:Gaurav Ghosal · Pratyush Maini · Aditi Raghunathan

Title: Disentangling Sequence Memorization and General Capability in LLMs

Abstract:
Large language models are susceptible to memorizing repeated sequences, posing serious privacy concerns. Existing attempts to isolate memorized information to specific neurons have seen limited success. Through experiments in a controlled setting, we demonstrate that standard training simply does not lead to ``memorization’’ neurons that can be easily isolated, unless the sequence is extremely atypical. Instead, we argue for a new training paradigm designed to simultaneously achieve two seemingly conflicting goals: isolation—confine memorization to specific neurons, and generalization—preserve learning across the dataset. Generalization requires a subset of shared neurons to be active across all sequences. Inspired by the dynamics of memorization, we argue that isolation can be encouraged even in the presence of shared neurons by simply activating a fixed prespecified set of memorization neurons for each sequence. Put together, we propose a scalable and practical method called sequence-tied dropout that enables perfect post-hoc isolation while preserving general model capabilities on the TinyStories dataset. To the best of our knowledge, this is the first proof-of-concept on small-scale real data showing that simultaneous generalization and isolation is in fact feasible.

Paperid:494

Authors:Tim Vieira · Benjamin LeBrun · Mario Giulianelli · Juan Luis Gastaldi · Brian DuSell · John Terilla · Timothy O'Donnell · Ryan Cotterell

Title: From Language Models over Tokens to Language Models over Characters

Abstract:
Modern language models are internally—and mathematically—distributions over token strings rather than character strings, posing numerous challenges for programmers building user applications on top of them. For example, if a prompt is specified as a character string, it must be tokenized before passing it to the tokenlevel language model. Thus, the tokenizer and consequent analyses are very sensitive to the specification of the prompt (e.g., if the prompt ends with a space or not). This paper presents algorithms for converting token-level language models to character-level ones. We present both exact and approximate algorithms. In the empirical portion of the paper, we benchmark the practical runtime and approximation quality. We find that---even with a small computation budget—our method is able to accurately approximate the character-level distribution (less than 0.00021 excess bits / character) at reasonably fast speeds (46.3 characters / second) on the Llama 3.1 8B language model.

Paperid:495

Authors:Emiliano Penaloza · Tianyue Zhang · Laurent Charlin · Mateo Espinosa Zarlenga

Title: Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization

Abstract:
Concept Bottleneck Models (CBMs) propose toenhance the trustworthiness of AI systems byconstraining their decisions on a set of humanunderstandable concepts. However, CBMs typ-ically rely on datasets with assumedly accurateconcept labels—an assumption often violated inpractice which we show can significantly degradeperformance. To address this, we introduce theConcept Preference Optimization (CPO) objec-tive, a new loss function based on Direct Pref-erence Optimization, which effectively mitigatesthe negative impact of concept mislabeling onCBM performance. We provide an analysis onsome key properties of the CPO objective show-ing it directly optimizes for the concept’s posteriordistribution, and contrast it against Binary CrossEntropy (BCE) where we show CPO is inherentlyless sensitive to concept noise. We empiricallyconfirm our analysis finding that CPO consistentlyoutperforms BCE in three real-world datasets withand without added label noise

Paperid:496

Authors:Nikolaos Chaidos · Angeliki Dimitriou · Maria Lymperaiou · Giorgos Stamou

Title: SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval

Abstract:
Despite the dominance of convolutional and transformerbased architectures in image-to-image retrieval, these models are prone to biases arising from low-level visual features, such as color. Recognizing the lack of semantic understanding as a key limitation, we propose a novel scene graph-based retrieval framework that emphasizes semantic content over superficial image characteristics. Prior approaches to scene graph retrieval predominantly rely on supervised Graph Neural Networks (GNNs), which require ground truth graph pairs driven from image captions. However, the inconsistency of caption-based supervision stemming from variable text encodings undermine retrieval reliability. To address these, we presentSCENIR, a Graph Autoencoder-based unsupervised retrieval framework, which eliminates the dependence on labeled training data. Our model demonstrates superior performance across metrics and runtime efficiency, outperforming existing vision-based, multimodal, and supervised GNN approaches. We further advocate forGraph Edit Distance(GED) as a deterministic and robust ground truth measure for scene graph similarity, replacing the inconsistent caption-based alternatives for the first time in image-to-image retrieval evaluation. Finally, we validate the generalizability of our method by applying it to unannotated datasets via automated scene graph generation, while substantially contributing in advancing state-of-the-art in counterfactual image retrieval.

Paperid:497

Authors:Yao Xiao · Binbin Yang · Weiyan Chen · Jiahao Chen · Zijie Cao · ZiYi Dong · Xiangyang Ji · Liang Lin · Wei Ke · Pengxu Wei

Title: Are High-Quality AI-Generated Images More Difficult for Models to Detect?

Abstract:
The remarkable evolution of generative models enables the generation of highquality and visually attractive images. Notably, these images can be perceptually indistinguishable from real photographs for human eyes, therefore attracting increasing attention to developing models for AI-generated image (AIGI) detection. Intuitively, as the visual quality of generated images improves, the difficulty of forgery detection increases accordingly. However, with a systematic study on four cutting-edge text-to-image generators, we intriguingly find that generated images with higher quality scores, as assessed by human preference models, tend to be more easily detected by existing AIGI detectors. To investigate this counterintuitive phenomenon, we study how the text condition provided to the generator and the characteristics of the generated images affect the quality scores and the accuracy of detectors. We observe that images generated from short prompts tend to achieve higher preference scores while being easier to detect. Moreover, we cluster the images based on their embeddings from preference models, and conduct regression analyses to verify that image characteristics, such as saturation, contrast, and texture richness, collectively influence both quality scores and AIGI detector accuracy. Finally, we showcase that the performance of the off-the-shelf detectors can be enhanced across diverse generators and datasets by selecting input patches based on the predicted scores of our regression models, thereby substantiating the broader applicability of our findings.

Paperid:498

Authors:Filippo Rinaldi · Giacomo Capitani · Lorenzo Bonicelli · Angelo Porrello · Donato Crisostomi · Federico Bolelli · Emanuele Rodola · ELISA FICARRA · Simone Calderara

Title: Update Your Transformer to the Latest Release: Re-Basin of Task Vectors

Abstract:
Foundation models serve as the backbone for numerous specialized models developed through finetuning. However, when the underlying pretrained model is updated or retrained (e.g., on larger and more curated datasets), the fine-tuned model becomes obsolete, losing its utility and requiring retraining. This raises the question: is it possible to transfer fine-tuning to a new release of the model? In this work, we investigate how to transfer fine-tuning to a new checkpoint without having to re-train, in a data-free manner. To do so, we draw principles from model re-basin and provide a recipe based on weight permutations to re-base the modifications made to the original base model, often called task vector. In particular, our approach tailors model re-basin for Transformer models, taking into account the challenges of residual connections and multi-head attention layers. Specifically, we propose a two-level method rooted in spectral theory, initially permuting the attention heads and subsequently adjusting parameters within select pairs of heads. Through extensive experiments on visual and textual tasks, we achieve the seamless transfer of fine-tuned knowledge to new pre-trained backbones without relying on a single training step or datapoint.

Paperid:499

Authors:Beepul Bharti · Mary Clemens-Sewall · Paul H. Yi · Jeremias Sulam

Title: Multiaccuracy and Multicalibration via Proxy Groups

Abstract:
As the use of predictive machine learning algorithms increases in highstakes decision-making, it is imperative that these algorithms are fair across sensitive groups. However, measuring and enforcing fairness in real-world applications can be challenging due to missing or incomplete sensitive group data. Proxy-sensitive attributes have been proposed as a practical and effective solution in these settings, but only for parity-based fairness notions. Knowing how to evaluate and control for fairness with missing sensitive group data for newer and more flexible frameworks, such as multiaccuracy and multicalibration, remains unexplored. In this work, we address this gap by demonstrating that in the absence of sensitive group data, proxy-sensitive attributes can provably be used to derive actionable upper bounds on the true multiaccuracy and multicalibration, providing insights into a model’s potential worst-case fairness violations. Additionally, we show that adjusting models to satisfy multiaccuracy and multicalibration across proxy-sensitive attributes can significantly mitigate these violations for the true, but unknown, demographic groups. Through several experiments on real-world datasets, we illustrate that approximate multiaccuracy and multicalibration can be achieved even when sensitive group data is incomplete or unavailable.

Paperid:500

Authors:Hong-You Chen · Zhengfeng Lai · Haotian Zhang · Xinze Wang · Marcin Eichner · Keen You · Meng Cao · Bowen Zhang · Yinfei Yang · Zhe Gan

Title: Contrastive Localized Language-Image Pre-Training

Abstract:
CLIP has been a celebrated method for training vision encoders to generate image/text representations facilitating various applications. Recently, it has been widely adopted as the vision backbone of multimodal large language models (MLLMs). The success of CLIP relies on aligning webcrawled noisy text annotations at image levels. However, such criteria may be insufficient for downstream tasks in need of fine-grained vision representations, especially when understanding region-level is demanding for MLLMs. We improve the localization capability of CLIP with several advances. Our proposed pre-training method, Contrastive Localized Language-Image Pre-training (CLOC), complements CLIP with region-text contrastive loss and modules. We formulate a new concept, promptable embeddings, of which the encoder produces image embeddings easy to transform into region representations given spatial hints. To support large-scale pre-training, we design a visually-enriched and spatially-localized captioning framework to effectively generate region-text labels. By scaling up to billions of annotated images, CLOC enables high-quality regional embeddings for recognition and retrieval tasks, and can be a drop-in replacement of CLIP to enhance MLLMs, especially on referring and grounding tasks.

Paperid:501

Authors:Amirhossein Roknilamouki · Arnob Ghosh · Ming Shi · Fatemeh Nourzad · Eylem Ekici · Ness Shroff

Title: Provably Efficient RL for Linear MDPs under Instantaneous Safety Constraints in Non-Convex Feature Spaces

Abstract:
Abstract:In Reinforcement Learning (RL), tasks with instantaneous hard constraints present significant challenges, particularly when the decision space is nonconvex or non-star-convex. This issue is especially relevant in domains like autonomous vehicles and robotics, where constraints such as collision avoidance often take a non-convex form. In this paper, we establish a regret bound of $\tilde{\mathcal{O}}\bigl(\bigl(1 + \tfrac{1}{\tau}\bigr) \sqrt{\log\bigl(\tfrac{1}{\tau}\bigr) d^3 H^4 K} \bigr)$, applicable to both star-convex and non-star-convex cases, where $d$ is the feature dimension, $H$ the episode length, $K$ the number of episodes, and $\tau$ the safety threshold. Moreover, the violation of safety constraints is zero with high probability throughout the learning process. A key technical challenge in these settings is bounding the covering number of the value-function class, which is essential for achieving value-aware uniform concentration in model-free function approximation. For the star-convex setting, we develop a novel technique called $\textit{Objective–Constraint Decomposition}$ (OCD) to properly bound the covering number. This result also resolves an error in a previous work on constrained RL. In non-star-convex scenarios, where the covering number can become infinitely large, we propose a two-phase algorithm, Non-Convex Safe Least Squares Value Iteration (NCS-LSVI), which first reduces uncertainty about the safe set by playing a known safe policy. After that, it carefully balances exploration and exploitation to achieve the regret bound. Finally, numerical simulations on an autonomous driving scenario demonstrate the effectiveness of NCS-LSVI.

Paperid:502

Authors:Avery Ma · Yangchen Pan · Amir-massoud Farahmand

Title: PANDAS: Improving Many-shot Jailbreaking via Positive Affirmation, Negative Demonstration, and Adaptive Sampling

Abstract:
Manyshot jailbreaking circumvents the safety alignment of large language models by exploiting their ability to process long input sequences. To achieve this, the malicious target prompt is prefixed with hundreds of fabricated conversational turns between the user and the model. These fabricated exchanges are randomly sampled from a pool of malicious questions and responses, making it appear as though the model has already complied with harmful instructions. In this paper, we present PANDAS: a hybrid technique that improves many-shot jailbreaking by modifying these fabricated dialogues with Positive Affirmations, Negative Demonstrations, and an optimized Adaptive Sampling method tailored to the target prompt's topic. Extensive experiments on AdvBench and HarmBench, using state-of-the-art LLMs, demonstrate that PANDAS significantly outperforms baseline methods in long-context scenarios. Through an attention analysis, we provide insights on how long-context vulnerabilities are exploited and show how PANDAS further improves upon many-shot jailbreaking.

Paperid:503

Authors:Gwanhyeong Koo · Sunjae Yoon · Younghwan Lee · Ji Woo Hong · Chang Yoo

Title: FlowDrag: 3D-aware Drag-based Image Editing with Mesh-guided Deformation Vector Flow Fields

Abstract:
Dragbased editing allows precise object manipulation through point-based control, offering user convenience. However, current methods often suffer from a geometric inconsistency problem by focusing exclusively on matching user-defined points, neglecting the broader geometry and leading to artifacts or unstable edits. We propose FlowDrag, which leverages geometric information for more accurate and coherent transformations. Our approach constructs a 3D mesh from the image, using an energy function to guide mesh deformation based on user-defined drag points. The resulting mesh displacements are projected into 2D and incorporated into a UNet denoising process, enabling precise handle-to-target point alignment while preserving structural integrity. Additionally, existing drag-editing benchmarks provide no ground truth, making it difficult to assess how accurately the edits match the intended transformations. To address this, we present VFD (VidFrameDrag) benchmark dataset, which provides ground-truth frames using consecutive shots in a video dataset. FlowDrag outperforms existing drag-based editing methods on both VFD Bench and DragBench.

Paperid:504

Authors:Shriya Bhatija · Paul-David Zuercher · Jakob Thumm · Thomas Bohné

Title: Multi-Objective Causal Bayesian Optimization

Abstract:
In decisionmaking problems, the outcome of an intervention often depends on the causal relationships between system components and is highly costly to evaluate. In such settings, causal Bayesian optimization (CBO) can exploit the causal relationships between the system variables and sequentially perform interventions to approach the optimum with minimal data. Extending CBO to handle multiple objectives, we proposemulti-objective causal Bayesian optimization(MO-CBO), a paradigm for identifying Pareto-optimal interventions within a known causal graph. We first derive a graphical characterization for potentially optimal sets of variables to intervene upon. Showing that any MO-CBO problem can be decomposed into several traditional multi-objective optimization tasks, we then introduce an algorithm that sequentially balances exploration across these tasks. The proposed method will be validated on both synthetic and real-world causal graphs, demonstrating its superiority over traditional (non-causal) multi-objective Bayesian optimization in settings where causal information is available.

Paperid:505

Authors:Kai Lu · Shixiong Xu · Jinqiu Li · Kun Ding · Gaofeng Meng

Title: Agent Reviewers: Domain-specific Multimodal Agents with Shared Memory for Paper Review

Abstract:
Feedback from peer review is essential to improve the quality of scientific articles. However, at present, many manuscripts do not receive sufficient external feedback for refinement before or during submission. Therefore, a system capable of providing detailed and professional feedback is crucial for enhancing research efficiency. In this paper, we have compiled the largest dataset of paper reviews to date by collecting historical openaccess papers and their corresponding review comments and standardizing them using LLM. We then developed a multi-agent system that mimics real human review processes, based on LLMs. This system, named Agent Reviewers, includes the innovative introduction of multimodal reviewers to provide feedback on the visual elements of papers. Additionally, a shared memory pool that stores historical papers' metadata is preserved, which supplies reviewer agents with background knowledge from different fields. Our system is evaluated using ICLR 2024 papers and achieves superior performance compared to existing AI-based review systems. Comprehensive ablation studies further demonstrate the effectiveness of each module and agent in this system.

Paperid:506

Authors:Xiaowen Ma · Zhen-Liang Ni · Shuai Xiao · Xinghao Chen

Title: TimePro: Efficient Multivariate Long-term Time Series Forecasting with Variable- and Time-Aware Hyper-state

Abstract:
In longterm time series forecasting, different variables often influence the target variable over distinct time intervals, a challenge known as the multi-delay issue. Traditional models typically process all variables or time points uniformly, which limits their ability to capture complex variable relationships and obtain non-trivial time representations. To address this issue, we propose TimePro, an innovative Mamba-based model that constructs variate- and time-aware hyper-states. Unlike conventional approaches that merely transfer plain states across variables or time dimension, TimePro preserves the fine-grained temporal features of each variate token and adaptively selects the focused time points to tune the plain state. The reconstructed hyper-state is able to perceive both variable relationship and salient temporal information, which helps the model make accurate forecasting. In experiments, TimePro achieves competitive performance on eight real-world long-term forecasting benchmarks with satisfactory linear complexity.

Paperid:507

Authors:Meshi Bashari · Matteo Sesia · Yaniv Romano

Title: Robust Conformal Outlier Detection under Contaminated Reference Data

Abstract:
Conformal prediction is a flexible framework for calibrating machine learning predictions, providing distributionfree statistical guarantees. In outlier detection, this calibration relies on a reference set of labeled inlier data to control the type-I error rate. However, obtaining a perfectly labeled inlier reference set is often unrealistic, and a more practical scenario involves access to a contaminated reference set containing a small fraction of outliers. This paper analyzes the impact of such contamination on the validity of conformal methods. We prove that under realistic, non-adversarial settings, calibration on contaminated data yields conservative type-I error control, shedding light on the inherent robustness of conformal methods. This conservativeness, however, typically results in a loss of power. To alleviate this limitation, we propose a novel, active data-cleaning framework that leverages a limited labeling budget and an outlier detection model to selectively annotate data points in the contaminated reference set that are suspected as outliers. By removing only the annotated outliers in this ``suspicious'' subset, we can effectively enhance power while mitigating the risk of inflating the type-I error rate, as supported by our theoretical analysis. Experiments on real datasets validate the conservative behavior of conformal methods under contamination and show that the proposed data-cleaning strategy improves power without sacrificing validity.

Paperid:508

Authors:Evan Sidrow · Alexandre Bouchard-Côté · Lloyd Elliott

Title: Variational Phylogenetic Inference with Products over Bipartitions

Abstract:
Bayesian phylogenetics requires accurate and efficient approximation of posterior distributions over trees. In this work, we develop a variational Bayesian approach for ultrametric phylogenetic trees. We present a novel variational family based on coalescent times of a singlelinkage clustering and derive a closed-form density of the resulting distribution over trees. Unlike existing methods for ultrametric trees, our method performs inference over all of tree space, it does not require any Markov chain Monte Carlo subroutines, and our variational family is differentiable. Through experiments on benchmark genomic datasets and an application to SARS-CoV-2, we demonstrate that our method achieves competitive accuracy while requiring significantly fewer gradient evaluations than existing state-of-the-art techniques.

Paperid:509

Authors:Xiong Wang · Yangze Li · Chaoyou Fu · Yike Zhang · Yunhang Shen · Lei Xie · Ke Li · Xing Sun · Long Ma

Title: Freeze-Omni: A Smart and Low Latency Speech-to-speech Dialogue Model with Frozen LLM

Abstract:
The GPT4o's excellent duplex speech interaction ability has given users an impressive experience. Researchers have recently proposed several multi-modal LLMs to achieve user-agent speech-to-speech conversations. In this paper, we propose a novel speech-text multimodal LLM architecture called Freeze-Omni, and our main contribution is that the speech input and output modalities can be easily connected to a textual LLM while keeping the LLM's parameters frozen throughout the training process. We effectively ensure that the intelligence of the Freeze-Omni in the speech modality is at the same level compared with that in the text modality of its backbone LLM, while achieving low latency end-to-end spoken response. In addition, we also designed a method to achieve duplex dialogue ability through multi-task training, giving Freeze-Omni a more natural style of dialogue ability between users and agents. In summary, Freeze-Omni holds great potential to conduct speech-to-speech dialogue based on a multimodal LLM under the condition of a frozen LLM, avoiding the catastrophic forgetting problem caused by limited data and training resources.

Paperid:510

Authors:Francesco Emanuele Stradi · Matteo Castiglioni · Alberto Marchesi · Nicola Gatti

Title: Learning Adversarial MDPs with Stochastic Hard Constraints

Abstract:
We study online learning in constrained Markov decision processes (CMDPs) with adversarial losses and stochastic hard constraints, under bandit feedback. We consider three scenarios. In the first one, we address general CMDPs, where we design an algorithm attaining sublinear regret and cumulative positive constraints violation. In the second scenario, under the mild assumption that a policy strictly satisfying the constraints exists and is known to the learner, we design an algorithm that achieves sublinear regret while ensuring that constraints are satisfied at every episode with high probability. In the last scenario, we only assume the existence of a strictly feasible policy, which is not known to the learner, and we design an algorithm attaining sublinear regret and constant cumulative positive constraints violation. Finally, we show that in the last two scenarios, a dependence on the Slater's parameter is unavoidable. To the best of our knowledge, our work is the first to study CMDPs involving both adversarial losses and hard constraints. Thus, our algorithms can deal with general nonstationary environments subject to requirements much stricter than those manageable with existing ones, enabling their adoption in a much wider range of applications.

Paperid:511

Authors:Oren Mangoubi · Neil He · Nisheeth K. Vishnoi

Title: Efficient Diffusion Models for Symmetric Manifolds

Abstract:
Abstract:We introduce a framework for designing efficient diffusion models for $d$dimensional symmetric-space Riemannian manifolds, including the torus, sphere, special orthogonal group and unitary group. Existing manifold diffusion models often depend on heat kernels, which lack closed-form expressions and require either $d$ gradient evaluations or exponential-in-$d$ arithmetic operations per training step. We introduce a new diffusion model for symmetric manifolds with a spatially-varying covariance, allowing us to leverage a projection of Euclidean Brownian motion to bypass heat kernel computations. Our training algorithm minimizes a novel efficient objective derived via Ito's Lemma, allowing each step to run in $O(1)$ gradient evaluations and nearly-linear-in-$d$ ($O(d^{1.19})$) *arithmetic* operations, reducing the gap between diffusions on symmetric manifolds and Euclidean space. Manifold symmetries ensure the diffusion satisfies an "average-case" Lipschitz condition, enabling accurate and efficient sample generation. Empirically, our model outperforms prior methods in training speed and improves sample quality on synthetic datasets on the torus, special orthogonal group, and unitary group.

Paperid:512

Authors:Xiulong Yuan · Hongtao Xu · Wenting Shen · Ang Wang · Xiafei Qiu · Jie Zhang · Yuqiong Liu · Bowen Yu · Junyang Lin · Mingzhen Li · Weile Jia · Yong Li · Wei Lin

Title: Efficient Long Context Fine-tuning with Chunk Flow

Abstract:
Long context finetuning of large language models(LLMs) involves training on datasets that are predominantly composed of short sequences and a small proportion of longer sequences. However, existing approaches overlook this long-tail distribution and employ training strategies designed specifically for long sequences. Moreover, these approaches also fail to address the challenges posed by variable sequence lengths during distributed training, such as load imbalance in data parallelism and severe pipeline bubbles in pipeline parallelism. These issues lead to suboptimal training performance and poor GPU resource utilization.To tackle these problems, we propose a chunk-centric training method named ChunkFlow. ChunkFlow reorganizes input sequences into uniformly sized chunks by consolidating short sequences and splitting longer ones. This approach achieves optimal computational efficiency and balance among training inputs. Additionally, ChunkFlow incorporates a state-aware chunk scheduling mechanism to ensure that the peak memory usage during training is primarily determined by the chunk size rather than the maximum sequence length in the dataset. Integrating this scheduling mechanism with existing pipeline scheduling algorithms further enhances the performance of distributed training.Experimental results demonstrate that, compared with Megatron-LM, ChunkFlow can be up to 4.53x faster in the long context fine-tuning of LLMs. Furthermore, we believe that ChunkFlow serves as an effective solution for a broader range of scenarios, such as long context continual pre-training, where datasets contain variable-length sequences.

Paperid:513

Authors:Chenglong Wang · Yang Gan · Yifu Huo · Yongyu Mu · Qiaozhi He · MuRun Yang · Bei Li · Tong Xiao · Chunliang Zhang · Tongran Liu · Jingbo Zhu

Title: GRAM: A Generative Foundation Reward Model for Reward Generalization

Abstract:
In aligning large language models (LLMs), reward models have played an important role, but are standardly trained as discriminative models and rely only on labeled human preference data. In this paper, we explore methods that train reward models using both unlabeled and labeled data. Building on the generative models in LLMs, we develop a generative reward model that is first trained via largescale unsupervised learning and then fine-tuned via supervised learning. We also show that by using label smoothing, we are in fact optimizing a regularized pairwise ranking loss. This result, in turn, provides a new view of training reward models, which links generative models and discriminative models under the same class of training objectives. The outcome of these techniques is a foundation reward model, which can be applied to a wide range of tasks with little or no further fine-tuning effort. Extensive experiments show that this model generalizes well across several tasks, including response ranking, reinforcement learning from human feedback, and task adaptation with fine-tuning, achieving significant performance improvements over several strong baseline models.

Paperid:514

Authors:Dimitri von Rütte · Janis Fluri · Yuhui Ding · Antonio Orvieto · Bernhard Schölkopf · Thomas Hofmann

Title: Generalized Interpolating Discrete Diffusion

Abstract:
While stateof-the-art language models achieve impressive results through next-token prediction, they have inherent limitations such as the inability to revise already generated tokens. This has prompted exploration of alternative approaches such as discrete diffusion. However, masked diffusion, which has emerged as a popular choice due to its simplicity and effectiveness, reintroduces this inability to revise words. To overcome this, we generalize masked diffusion and derive the theoretical backbone of a family of general interpolating discrete diffusion (GIDD) processes offering greater flexibility in the design of the noising processes. Leveraging a novel diffusion ELBO, we achieve compute-matched state-of-the-art performance in diffusion language modeling. Exploiting GIDD's flexibility, we explore a hybrid approach combining masking and uniform noise, leading to improved sample quality and unlocking the ability for the model to correct its own mistakes, an area where autoregressive models notoriously have struggled. Our code and models will be released upon acceptance.

Paperid:515

Authors:Shenao Zhang · Zhihan Liu · Boyi Liu · Yufeng Zhang · Yingxiang Yang · Yongfei Liu · Liyu Chen · Tao Sun · Zhaoran Wang

Title: Reward-Augmented Data Enhances Direct Preference Alignment of LLMs

Abstract:
Preference alignment in Large Language Models (LLMs) has significantly improved their ability to adhere to human instructions and intentions. However, existing direct alignment algorithms primarily focus on relative preferences and often overlook the qualitative aspects of responses, despite having access to preference data that includes reward scores from judge models during AI feedback. Striving to maximize the implicit reward gap between the chosen and the slightly inferior rejected responses can cause overfitting and unnecessary unlearning of the highquality rejected responses. The unawareness of the reward scores also drives the LLM to indiscriminately favor the low-quality chosen responses and fail to generalize to optimal responses that are sparse in data. To overcome these shortcomings, our study introduces reward-conditioned LLM policies that discern and learn from the entire spectrum of response quality within the dataset, helping extrapolate to more optimal regions. We propose an effective yet simple data relabeling method that conditions the preference pairs on quality scores to construct a reward-augmented dataset. The experimental results across various instruction-following and academic benchmarks demonstrate that our approach consistently boosts the performance of DPO by a considerable margin across diverse models. When applying our method to on-policy data, the resulting DPO model outperforms various baselines and achieves state-of-the-art results. Through comprehensive ablation studies, we demonstrate that our method not only maximizes the utility of preference data but also mitigates the issue of unlearning, demonstrating its broad effectiveness beyond mere data expansion.

Paperid:516

Authors:Yuanze Wang · Yichao Yan · Jin · Shiming Song · Yilan Huang · Xingdong Sheng · Dianxi Shi

Title: Enhancing Visual Localization with Cross-Domain Image Generation

Abstract:
Visual localization aims to predict the absolute camera pose for a single query image. However, predominant methods focus on singlecamera images and scenes with limited appearance variations, limiting their applicability to cross-domain scenes commonly encountered in real-world applications. Furthermore, the long-tail distribution of cross-domain datasets poses additional challenges for visual localization. In this work, we propose a novel cross-domain data generation framework to enhance visual localization methods. To achieve this, we first construct a cross-domain 3DGS to accurately model photometric variations and mitigate the interference of dynamic objects in large-scale scenes. We introduce a text-guided image editing model to enhance data diversity for addressing the long-tail distribution problem and design an effective fine-tuning strategy for it. Then, we develop an anchor-based method to generate high-quality datasets for visual localization. Finally, we introduce positional attention to address data ambiguities in cross-camera images. Extensive experiments show that our method achieves state-of-the-art accuracy, outperforming existing cross-domain visual localization methods by an average of 59% across all domains.

Paperid:517

Authors:Yuhang Cai · Kangjie Zhou · Jingfeng Wu · Song Mei · Michael Lindsey · Peter Bartlett

Title: Implicit Bias of Gradient Descent for Non-Homogeneous Deep Networks

Abstract:
We establish the asymptotic implicit bias of gradient descent (GD) for generic nonhomogeneous deep networks under exponential loss. Specifically, we characterize three key properties of GD iterates starting from a sufficiently small empirical risk, where the threshold is determined by a measure of the network's non-homogeneity. First, we show that a normalized margin induced by the GD iterates increases nearly monotonically. Second, we prove that while the norm of the GD iterates diverges to infinity, the iterates themselves converge in direction. Finally, we establish that this directional limit satisfies the Karush–Kuhn–Tucker (KKT) conditions of a margin maximization problem. Prior works on implicit bias have focused exclusively on homogeneous networks; in contrast, our results apply to a broad class of non-homogeneous networks satisfying a mild near-homogeneity condition. In particular, our results apply to networks with residual connections and non-homogeneous activation functions, thereby resolving an open problem posed by \citet{ji2020directional}.

Paperid:518

Authors:Zixiang Ai · Zichen Liu · Jiahuan Zhou

Title: Vision Graph Prompting via Semantic Low-Rank Decomposition

Abstract:
Vision GNN (ViG) demonstrates superior performance by representing images as graph structures, providing a more natural way to capture irregular semantic patterns beyond traditional grid or sequencebased representations. To efficiently adapt ViG to downstream tasks, parameter-efficient fine-tuning techniques like visual prompting become increasingly essential. However, existing prompting methods are primarily designed for Transformer-based models, neglecting the rich topological relationships among nodes and edges in graph-based representations, limiting their capacity to model complex semantics. In this paper, we propose Vision Graph Prompting (VGP), a novel framework tailored for vision graph structures. Our core insight reveals that semantically connected components in the graph exhibit low-rank properties. Building on this observation, we introduce a semantic low-rank prompting method that decomposes low-rank semantic features and integrates them with prompts on vision graph topologies, capturing both global structural patterns and fine-grained semantic dependencies. Extensive experiments demonstrate our method significantly improves ViG’s transfer performance on diverse downstream tasks, achieving results comparable to full fine-tuning while maintaining parameter efficiency.

Paperid:519

Authors:Michael Crawshaw · Blake Woodworth · Mingrui Liu

Title: Constant Stepsize Local GD for Logistic Regression: Acceleration by Instability

Abstract:
Abstract:Existing analysis of Local (Stochastic) Gradient Descent for heterogeneous objectives requires stepsizes $\eta \leq 1/K$ where $K$ is the communication interval, which ensures monotonic decrease of the objective. In contrast, we analyze Local Gradient Descent for logistic regression with separable, heterogeneous data using any stepsize $\eta > 0$. With $R$ communication rounds and $M$ clients, we show convergence at a rate $\mathcal{O}(1/\eta K R)$ after an initial unstable phase lasting for $\widetilde{\mathcal{O}}(\eta K M)$ rounds. This improves upon the existing $\mathcal{O}(1/R)$ rate for general smooth, convex objectives. Our analysis parallels the single machine analysis of Wu et al. (2024) in which instability is caused by extremely large stepsizes, but in our setting another source of instability is large local updates with heterogeneous objectives.

Paperid:520

Authors:Haohan Zou · Jie Feng · Hao Zhao · Yuanyuan Shi

Title: Analytical Lyapunov Function Discovery: An RLbased Generative Approach

Abstract:
Despite advances in learningbased methods, finding valid Lyapunov functions for nonlinear dynamical systems remains challenging. Current neural network approaches face two main issues: challenges in scalable verification and limited interpretability. To address these, we propose an end-to-end framework using transformers to construct analytical Lyapunov functions (local), which simplifies formal verification, enhances interpretability, and provides valuable insights for control engineers. Our framework consists of a transformer-based trainer that generates candidate Lyapunov functions and a falsifier that verifies candidate expressions and refines the model via risk-seeking policy gradient. Unlike Alfarano et al. (2024), which utilizes pre-training and seeks global Lyapunov functions for low-dimensional systems, our model is trained from scratch via reinforcement learning and succeeds in finding local Lyapunov functions for high-dimensional and non-polynomial systems. Given the analytical nature of the candidates, we employ efficient optimization methods for falsification during training and formal verification tools for the final verification. We demonstrate the efficiency of our approach on a range of nonlinear dynamical systems with up to ten dimensions and show that it can discover Lyapunov functions not previously identified in the control literature.

Paperid:521

Authors:Julia Gusak · Xunyi Zhao · Théotime Le Hellard · Zhe LI · Lionel Eyraud-Dubois · Olivier Beaumont

Title: HiRemate: Hierarchical Approach for Efficient Re-materialization of Neural Networks

Abstract:
Training modern neural networks on memorylimited GPUs is challenging, as storing intermediate activations often exceeds available memory. Re-materialization, a technique that preserves exact computations, addresses this by selectively recomputing activations instead of storing them. However, existing methods either fail to scale, lack generality, or introduce excessive execution overhead. We introduce HiRemate, a hierarchical re-materialization framework that recursively partitions large computation graphs, applies optimized solvers at multiple levels, and merges solutions into a global efficient training schedule. This enables scalability to significantly larger graphs than prior ILP-based methods while keeping runtime overhead low. Designed for single-GPU models and activation re-materialization, HiRemate extends the feasibility of training networks with thousands of graph nodes, surpassing prior methods in both efficiency and scalability. Experiments on various types of networks yield up to 50-70% memory reduction with only 10-15% overhead, closely matching optimal solutions while significantly reducing solver time. Seamlessly integrating with PyTorch Autograd, HiRemate requires almost no code change to use, and will be open-sourced, enabling broad adoption in memory-constrained deep learning.

Paperid:522

Authors:Ningyuan Huang · Miguel Sarabia · Abhinav Moudgil · Pau Rodriguez · Luca Zappella · Federico Danieli

Title: Understanding Input Selectivity in Mamba: Impact on Approximation Power, Memorization, and Associative Recall Capacity

Abstract:
StateSpace Models (SSMs), and particularly Mamba, have recently emerged as a promising alternative to Transformers. Mamba introduces input selectivity to its SSM layer (S6) and incorporates convolution and gating into its Mamba block. While these modifications do improve Mamba's performance over its SSM predecessors, it remains largely unclear how Mamba leverages the additional functionalities provided by input selectivity, and how these interact with the other operations in the Mamba architecture. In this work, we demystify the role of input selectivity in Mamba, investigating its impact on function approximation power, long-term memory, and associative recall capabilities.In particular: (i) We prove that the S6 layer of Mamba can represent projections ontoHaar wavelets, providing an edge over the S4D layer in approximating discontinuous functions commonly arising in practice. (ii) We show how the S6 layer suffers from memory decay, and describe how it can dynamically counteract it. (iii) We provide analytical solutions to the MQAR associative recall task using the Mamba architecture with different mixers --- Mamba, Mamba-2, and S4D. We demonstrate the tightness of our theoretical constructions with empirical results on concrete tasks. Our findings offer a mechanistic understanding of Mamba and reveal opportunities for improvement.

Paperid:523

Authors:Peng Wang · Yong Li · Xiu-Shen Wei · Lin Zhao

Title: Learning Attribute-Aware Hash Codes for Fine-Grained Image Retrieval via Query Optimization

Abstract:
Finegrained hashing has become a powerful solution for rapid and efficient image retrieval, particularly in scenarios requiring high discrimination between visually similar categories. To enable each hash bit to correspond to specific visual attributes, we propose a novel method that harnesses learnable queries for attribute-aware hash code learning. This method deploys a tailored set of queries to capture and represent nuanced attribute-level information within the hashing process, thereby enhancing both the interpretability and relevance of each hash bit. Building on this query-based optimization framework, we incorporate an auxiliary branch to help alleviate the challenges of complex landscape optimization often encountered with low-bit hash codes. This auxiliary branch models high-order attribute interactions, reinforcing the robustness and specificity of the generated hash codes. Experimental results on benchmark datasets demonstrate that our method generates attribute-aware hash codes and consistently outperforms state-of-the-art techniques in retrieval accuracy and robustness, especially for low-bit hash codes, underscoring its potential in fine-grained image hashing tasks.

Paperid:524

Authors:Weihua Du · Yiming Yang · Sean Welleck

Title: Optimizing Temperature for Language Models with Multi-Sample Inference

Abstract:
Multisample aggregation strategies, such as majority voting and best-of-N sampling, are widely used in contemporary large language models (LLMs) to enhance predictive accuracy across various tasks. A key challenge in this process is temperature selection, which significantly impacts model performance. Existing approaches either rely on a fixed default temperature or require labeled validation data for tuning, which are often scarce and difficult to obtain. This paper addresses the challenge of automatically identifying the (near)-optimal temperature for different LLMs using multi-sample aggregation strategies, without relying on task-specific validation data. We provide a comprehensive analysis of temperature’s role in performance optimization, considering variations in model architectures, datasets, task types, model sizes, and predictive accuracy. Furthermore, we propose a novel entropy-based metric for automated temperature optimization, which consistently outperforms fixed-temperature baselines. Additionally, we incorporate a stochastic process model to enhance interpretability, offering deeper insights into the relationship between temperature and model performance.

Paperid:525

Authors:Shyam Nuggehalli · Jifan Zhang · Lalit Jain · Robert Nowak

Title: Improved Algorithm for Deep Active Learning under Imbalance via Optimal Separation

Abstract:
Class imbalance severely impacts machine learning performance on minority classes in realworld applications. While various solutions exist, active learning offers a fundamental fix by strategically collecting balanced, informative labeled examples from abundant unlabeled data. We introduce DIRECT, an algorithm that identifies class separation boundaries and selects the most uncertain nearby examples for annotation. By reducing the problem to one-dimensional active learning, DIRECT leverages established theory to handle batch labeling and label noise -- another common challenge in data annotation that particularly affects active learning methods.Our work presents the first comprehensive study of active learning under both class imbalance and label noise. Extensive experiments on imbalanced datasets show DIRECT reduces annotation costs by over 60\% compared to state-of-the-art active learning methods and over 80\% versus random sampling, while maintaining robustness to label noise.

Paperid:526

Authors:Peijie Dong · Zhenheng Tang · Xiang Liu · Lujun Li · Xiaowen Chu · Bo Li

Title: Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compression

Abstract:
Posttraining compression reduces the computational and memory costs of large language models (LLMs), enabling resource-efficient deployment. However, existing compression benchmarks focus narrowly on language modeling (e.g., perplexity) and natural language understanding tasks (e.g., GLUE accuracy), ignoring the agentic capabilities—workflow, tool use/function call, long-context understanding and real-world application. We introduce the Agent Compression Benchmark (ACBench), the first comprehensive benchmark for evaluating how compression impacts LLMs' agentic abilities. ACBench spans (1) 12 tasks across 4 capabilities (e.g., WorfBench for workflow generation, Needle-in-Haystack for long-context retrieval), (2) 4-bit quantization (GPTQ, AWQ) and 50% pruning (Wanda, SparseGPT), and (3) 15 models, including small (Gemma-2B), standard (Qwen2.5-7B), and distilled reasoning LLMs (DeepSeek-R1-Distill). Our experiments reveal compression tradeoffs: 4-bit quantization preserves workflow generation and tool use (1%--3% drop) but degrades real-world application accuracy by 10%--15%. We introduce ERank, Top-k Ranking Correlation and Energy to systematize analysis. ACBench provides actionable insights for optimizing LLM compression in agentic scenarios, bridging the gap between algorithmic efficiency and real-world applicability.

Paperid:527

Authors:Daniel Csillag · Claudio Struchiner · Guilherme Tegoni Goedert

Title: Prediction-Powered E-Values

Abstract:
Quality statistical inference requires a sufficient amount of data, which can be missing or hard to obtain. To this end, predictionpowered inference has risen as a promising methodology, but existing approaches are largely limited to Z-estimation problems such as inference of means and quantiles. In this paper, we apply ideas of prediction-powered inference to e-values. By doing so, we inherit all the usual benefits of e-values -- such as anytime-validity, post-hoc validity and versatile sequential inference -- as well as greatly expand the set of inferences achievable in a prediction-powered manner. In particular, we show that every inference procedure that can be framed in terms of e-values has a prediction-powered counterpart, given by our method. We showcase the effectiveness of our framework across a wide range of inference tasks, from simple hypothesis testing and confidence intervals to more involved procedures for change-point detection and causal discovery, which were out of reach of previous techniques. Our approach is modular and easily integrable into existing algorithms, making it a compelling choice for practical applications.

Paperid:528

Authors:Qiyao Liang · Daoyuan Qian · Liu Ziyin · Ila R. Fiete

Title: Compositional Generalization Requires More Than Disentangled Representations

Abstract:
Composition—the ability to generate myriad variations from finite means—is believed to underlie powerful generalization. However, compositional generalization remains a key challenge for deep learning. A widely held assumption is that learning disentangled (factorized) representations naturally supports this kind of extrapolation. Yet, empirical results are mixed, with many generative models failing to recognize and compose factors to generate outof-distribution (OOD) samples. In this work, we investigate a controlled 2D Gaussian "bump" generation task, demonstrating that standard generative architectures fail in OOD regions when training with partial data, even when supplied with fully disentangled (x,y) coordinates, re-entangling them through subsequent layers. By examining the model's learned kernels and manifold geometry, we show that this failure reflects a "memorization" strategy for generation through the superposition of training data rather than by combining the true factorized features. We show that models forced—through architectural modifications with regularization or curated training data—to create disentangled representations in the full-dimensional representational (pixel) space can be highly data-efficient and effective at learning to compose in OOD regions. These findings underscore that bottlenecks with factorized/disentangled representations in an abstract representation are insufficient: the model must actively maintain or induce factorization directly in the representational space in order to achieve robust compositional generalization.

Paperid:529

Authors:Scott Emmons · Caspar Oesterheld · Vincent Conitzer · Stuart Russell

Title: Observation Interference in Partially Observable Assistance Games

Abstract:
We study partially observable assistance games (POAGs), a model of the humanAI value alignment problem which allows the human and the AI assistant to have partial observations. Motivated by concerns of AI deception, we study a qualitatively new phenomenon made possible by partial observability: would an AI assistant ever have an incentive to interfere with the human's observations? First, we prove that sometimes an optimal assistant must take observation-interferingactions, even when the human is playing optimally, and even when there are otherwise-equivalent actions available that do not interfere with observations. Though this result seems to contradict the classic theorem from single-agent decision making that the value of perfect information is nonnegative, we resolve this seeming contradiction by developing a notion of interference defined on entirepolicies. This can be viewed as an extension of the classic result that the value of perfect information is nonnegative into the cooperative multiagent setting. Second, we prove that if the human is simply making decisions based on their immediate outcomes, the assistant might need to interfere with observations as a way to query the human's preferences. We show that this incentive for interference goes away if the human is playing optimally, or if we introduce a communication channel for the human to communicate their preferences to the assistant. Third, we show that if the human acts according to the Boltzmann model of irrationality, this can create an incentive for the assistant to interfere with observations. Finally, we use an experimental model to analyze tradeoffs faced by the AI assistant in practice when considering whether or not to take observation-interfering actions.

Paperid:530

Authors:Jianze Li · Jiezhang Cao · Yong Guo · Wenbo Li · Yulun Zhang

Title: One Diffusion Step to Real-World Super-Resolution via Flow Trajectory Distillation

Abstract:
Diffusion models (DMs) have significantly advanced the development of realworld image super-resolution (Real-ISR), but the computational cost of multi-step diffusion models limits their application. One-step diffusion models generate high-quality images in a one sampling step, greatly reducing computational overhead and inference latency. However, most existing one-step diffusion methods are constrained by the performance of the teacher model, where poor teacher performance results in image artifacts. To address this limitation, we propose FluxSR, a novel one-step diffusion Real-ISR technique based on flow matching models. We use the state-of-the-art diffusion model FLUX.1-dev as both the teacher model and the base model. First, we introduce Flow Trajectory Distillation (FTD) to distill a multi-step flow matching model into a one-step Real-ISR. Second, to improve image realism and address high-frequency artifact issues in generated images, we propose TV-LPIPS as a perceptual loss and introduce Attention Diversification Loss (ADL) as a regularization term to reduce token similarity in transformer, thereby eliminating high-frequency artifacts. Comprehensive experiments demonstrate that our method outperforms existing one-step diffusion-based Real-ISR methods. The code and models will be available.

Paperid:531

Authors:Jan Pauls · Max Zimmer · Berkant Turan · Sassan Saatchi · Philippe CIAIS · Sebastian Pokutta · Fabian Gieseke

Title: Capturing Temporal Dynamics in Large-Scale Tree Canopy Height Estimation

Abstract:
With the rise in global greenhouse gas emissions, accurate largescale tree canopy height maps are essential for understanding forest structure, estimating above-ground biomass, and monitoring ecological disruptions. To this end, we present a novel approach to generate large-scale, high-resolution canopy height maps over time. Our model accurately predicts canopy height over multiple years given Sentinel 2 time series satellite data. Using GEDI LiDAR data as the ground truth for training the model, we present the first 10 m resolution temporal canopy height map of the European continent for the period 2019–2022. As part of this product, we also offer a detailed canopy height map for 2020, providing more precise estimates than previous studies. Our pipeline and the resulting temporal height map are publicly available, enabling comprehensive large-scale monitoring of forests and, hence, facilitating future research and ecological analyses. For an interactive viewer, see https://europetreemap.projects.earthengine.app/view/europeheight.

Paperid:532

Authors:Shuting He · Guangquan Jie · Changshuo Wang · Yun Zhou · Shuming Hu · Guanbin Li · Henghui Ding

Title: ReferSplat: Referring Segmentation in 3D Gaussian Splatting

Abstract:
We introduce Referring 3D Gaussian Splatting Segmentation (R3DGS), a new task that focuses on segmenting target objects in a 3D Gaussian scene based on natural language descriptions. This task requires the model to identify newly described objects that may be occluded or not directly visible in a novel view, posing a significant challenge for 3D multimodal understanding. Developing this capability is crucial for advancing embodied AI. To support research in this area, we construct the first R3DGS dataset, Ref-LERF. Our analysis reveals that 3D multi-modal understanding and spatial relationship modeling are key challenges for R3DGS. To address these challenges, we propose ReferSplat, a framework that explicitly models 3D Gaussian points with natural language expressions in a spatially aware paradigm. ReferSplat achieves state-of-the-art performance on both the newly proposed R3DGS task and 3D open-vocabulary segmentation benchmarks. Code, trained models, and the dataset will be publicly released.

Paperid:533

Authors:Simon Park · Abhishek Panigrahi · Yun Cheng · Dingli Yu · Anirudh Goyal · Sanjeev Arora

Title: Generalizing from SIMPLE to HARD Visual Reasoning: Can We Mitigate Modality Imbalance in VLMs?

Abstract:
Vision Language Models (VLMs) are impressive at visual question answering and image captioning. But they underperform on multistep visual reasoning---even compared to LLMs on the same tasks presented in text form---giving rise to perceptions ofmodality imbalanceorbrittleness.Towards a systematic study of such issues, we introduce a synthetic framework for assessing the ability of VLMs to perform algorithmic visual reasoning, comprising three tasks: Table Readout, Grid Navigation, and Visual Analogy. Each has two levels of difficulty, SIMPLE and HARD, and even the SIMPLE versions are difficult for frontier VLMs. We propose strategies for training on the SIMPLE version of tasks that improve performance on the corresponding HARD task, i.e., simple-to-hard (S2H) generalization. This controlled setup, where each task also has an equivalent text-only version, allows a quantification of the modality imbalance and how it is impacted by training strategy. We show that 1) explicit image-to-text conversion is important in promoting S2H generalization on images, by transferring reasoning from text; 2) conversion can be internalized at test time. We also report results of mechanistic study of this phenomenon. We identify measures of gradient alignment that can identify training strategies that promote better S2H generalization. Ablations highlight the importance of chain-of-thought.

Paperid:534

Authors:Ying Nie · Binwei Yan · Tianyu Guo · Zheyuan Bai · Mengyu Zheng · Hanting Chen · Jing Han

Title: MoRAgent: Parameter Efficient Agent Tuning with Mixture-of-Roles

Abstract:
Despite recent advancements of finetuning open-source large language models (LLMs) to facilitate agent tasks, parameter-efficient fine-tuning (PEFT) methodologies for agent remain largely unexplored. In this paper, we introduce three key strategies for PEFT in agent tasks: 1) Inspired by the increasingly dominant Reason+Action paradigm, we first decompose the capabilities necessary for the agent tasks into three distinct roles: reasoner, executor, and summarizer. The reasoner is responsible for comprehending the user's query and determining the next role based on the execution trajectory. The executor is tasked with identifying the appropriate functions and parameters to invoke. The summarizer conveys the distilled information from conversations back to the user. 2) We then propose the Mixture-of-Roles (MoR) framework, which comprises three specialized Low-Rank Adaptation (LoRA) groups, each designated to fulfill a distinct role. By focusing on their respective specialized capabilities and engaging in collaborative interactions, these LoRAs collectively accomplish the overall agent task. 3) To effectively fine-tune the framework, we develop a multi-role data generation pipeline based on publicly available datasets, incorporating role-specific content completion and reliability verification.We conduct extensive experiments and thorough ablation studies on various LLMs and agent benchmarks, demonstrating the effectiveness of the proposed method.

Paperid:535

Authors:Yi-Fan Zhang · Min-Ling Zhang

Title: Tight and Fast Bounds for Multi-Label Learning

Abstract:
Commonly used evaluation metrics in multilabel learning all involve base loss functions, and the theoretical guarantees of multi-label learning often rely on the properties of base loss functions. Some recent theoretical works have used the Lipschitz continuity of base loss functions to prove the generalization bounds for multi-label learning, but the impact of the smoothness of base loss functions on the generalization bounds is completely unknown. In an attempt to make up for this gap in the generalization theory of multi-label learning, we develop some novel vector-contraction inequalities for smooth base loss functions and derive tight generalization bounds with no dependency on the number of labels, up to logarithmic terms. We then exploit local Rademacher complexity to develop some novel local vector-contraction inequalities for smooth base loss functions, which induce generalization bounds with a tighter dependency on the number of labels and a faster convergence rate with respect to the number of examples. In addition, we derive tight generalization bounds with no dependency on the number of labels, up to logarithmic terms, for Macro-Averaged AUC by exploiting the Lipschitz continuity and smoothness of base loss functions, respectively. Our state-of-the-art theoretical results provide general theoretical guarantees for the generalization of multi-label learning.

Paperid:536

Authors:Lukas Lao Beyer · Tianhong Li · Xinlei Chen · Sertac Karaman · Kaiming He

Title: Highly Compressed Tokenizer Can Generate Without Training

Abstract:
Commonly used image tokenizers produce a 2D grid of spatially arranged tokens. In contrast, socalled1Dimage tokenizers represent images as highly compressed one-dimensional sequences of as few as 32 discrete tokens. We find that the high degree of compression achieved by a 1D tokenizer with vector quantization enables image editing and generative capabilities through heuristic manipulation of tokens, demonstrating that even very crude manipulations -- such as copying and replacing tokens between latent representations of images -- enable fine-grained image editing by transferring appearance and semantic attributes. Motivated by the expressivity of the 1D tokenizer's latent space, we construct an image generation pipeline leveraging gradient-based test-time optimization of tokens with plug-and-play loss functions such as reconstruction or CLIP similarity. Our approach is demonstrated for inpainting and text-guided image editing use cases, and can generate diverse and realistic samples without requiring training of any generative model.

Paperid:537

Authors:Ziang Chen · Xiaohan Chen · Jialin Liu · Xinshang Wang · Wotao Yin

Title: Expressive Power of Graph Neural Networks for (Mixed-Integer) Quadratic Programs

Abstract:
Quadratic programming (QP) is the most widely applied category of problems in nonlinear programming. Many applications require realtime/fast solutions, though not necessarily with high precision. Existing methods either involve matrix decomposition or use the preconditioned conjugate gradient method. For relatively large instances, these methods cannot achieve the real-time requirement unless there is an effective preconditioner. Recently, graph neural networks (GNNs) opened new possibilities for QP. Some promising empirical studies of applying GNNs for QP tasks show that GNNs can capture key characteristics of an optimization instance and provide adaptive guidance accordingly to crucial configurations during the solving process, or directly provide an approximate solution. However, the theoretical understanding of GNNs in this context remains limited. Specifically, it is unclear what GNNs can and cannot achieve for QP tasks in theory. This work addresses this gap in the context of linearly constrained QP tasks. In the continuous setting, we prove that message-passing GNNs can universally represent fundamental properties of quadratic programs, including feasibility, optimal objective values, and optimal solutions. In the more challenging mixed-integer setting, while GNNs are not universal approximators, we identify a subclass of QP problems that GNNs can reliably represent.

Paperid:538

Authors:Christopher Scarvelis · David Benhaim · Paul Zhang

Title: Orient Anything

Abstract:
Orientation estimation is a fundamental task in 3D shape analysis which consists of estimating a shape's orientation axes: its side, up-, and front-axes. Using this data, one can rotate a shape into canonical orientation, where its orientation axes are aligned with the coordinate axes. Developing an orientation algorithm that reliably estimates complete orientations of general shapes remains an open problem. We introduce a two-stage orientation pipeline that achieves state of the art performance on up-axis estimation and further demonstrate its efficacy on full-orientation estimation, where one seeks all three orientation axes. Unlike previous work, we train and evaluate our method on all of Shapenet rather than a subset of classes. We motivate our engineering contributions by theory describing fundamental obstacles to orientation estimation for rotationally-symmetric shapes, and show how our method avoids these obstacles.

Paperid:539

Authors:Jongwoo Ko · Tianyi Chen · Sungnyun Kim · Tianyu Ding · Luming Liang · Ilya Zharkov · Se-Young Yun

Title: DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMs

Abstract:
Despite the success of distillation in large language models (LLMs), most prior work applies identical loss functions to both teacherand student-generated data. These strategies overlook the synergy between loss formulations and data types, leading to a suboptimal performance boost in student models. To address this, we propose DistiLLM-2, a contrastive approach that simultaneously increases the likelihood of teacher responses and decreases that of student responses by harnessing this synergy. Our extensive experiments show that DistiLLM-2 not only builds high-performing student models across a wide range of tasks, including instruction-following and code generation, but also supports diverse applications, such as preference alignment and vision-language extensions. These findings highlight the potential of a contrastive approach to enhance the efficacy of LLM distillation by effectively aligning teacher and student models across varied data types.

Paperid:540

Authors:Tom A. Lamb · Adam Davies · Alasdair J Paren · Phil Torr · Francesco Pinto

Title: Focus On This, Not That! Steering LLMs With Adaptive Feature Specification

Abstract:
Despite the success of Instruction Tuning (IT) in training large language models (LLMs) to perform arbitrary userspecified tasks, these models often still leverage spurious or biased features learned from their training data, leading to undesired behaviours when deploying them in new contexts. In this work, we introduce Focus Instruction Tuning (FIT), which trains LLMs to condition their responses by focusing on specific features whilst ignoring others, leading to different behaviours based on what features are specified. Across several experimental settings, we show that focus-tuned models can be adaptively steered by focusing on different features at inference-time: for instance, robustness can be improved by focusing on core task features and ignoring spurious features, and social bias can be mitigated by ignoring demographic categories. Furthermore, FIT can steer behaviour in new contexts, generalising under distribution shift and to new unseen features at inference time, and thereby facilitating more robust, fair, and controllable LLM applications in real-world environments.

Paperid:541

Authors:Edith Cohen · Mihir Singhal · Uri Stemmer

Title: Breaking the Quadratic Barrier: Robust Cardinality Sketches for Adaptive Queries

Abstract:
Abstract:Cardinality sketches are compact data structures that efficiently estimate the number of distinct elements across multiple queries while minimizing storage, communication, and computational costs. However, recent research has shown that these sketches can fail under {\em adaptively chosen queries}, breaking down after approximately $\tilde{O}(k^2)$ queries, where $k$ is the sketch size.In this work, we overcome this \emph{quadratic barrier} by designing robust estimators with finegrained guarantees. Specifically, our constructions can handle an {\em exponential number of adaptive queries}, provided that each element participates in at most $\tilde{O}(k^2)$ queries. This effectively shifts the quadratic barrier from the total number of queries to the number of queries {\em sharing the same element}, which can be significantly smaller. Beyond cardinality sketches, our approach expands the toolkit for robust algorithm design.

Paperid:542

Authors:chao ying · Jun Jin · Yi Guo · Xiudi Li · Muxuan Liang · Jiwei Zhao

Title: Towards the Efficient Inference by Incorporating Automated Computational Phenotypes under Covariate Shift

Abstract:
Collecting goldstandard phenotype data via manual extraction is typically labor-intensive and slow, whereas automated computational phenotypes (ACPs) offer a systematic and much faster alternative.However, simply replacing the gold-standard with ACPs, without acknowledging their differences, could lead to biased results and misleading conclusions.Motivated by the complexity of incorporating ACPs while maintaining the validity of downstream analyses, in this paper, we consider a semi-supervised learning setting that consists of both labeled data (with gold-standard) and unlabeled data (without gold-standard), under the covariate shift framework.We develop doubly robust and semiparametrically efficient estimators that leverage ACPs for general target parameters in the unlabeled and combined populations. In addition,we carefully analyze the efficiency gains achieved by incorporating ACPs, comparing scenarios with and without their inclusion.Notably, we identify that ACPs for the unlabeled data, instead of for the labeled data, drive the enhanced efficiency gains. To validate our theoretical findings, we conduct extensive synthetic experiments and apply our method to a real-world dataset, confirming the practical advantages of our approach.

Paperid:543

Authors:Jiujiang Guo · Mankun Zhao · Wenbin Zhang · Tianyi Xu · Linying Xu · Yu Jian · Yu Mei · Yu Ruiguo

Title: TeDS: Joint Learning of Diachronic and Synchronic Perspectives in Quaternion Space for Temporal Knowledge Graph Completion

Abstract:
Existing research on temporal knowledge graph completion treats temporal information as supplementary, without simulating various features of facts from a temporal perspective. This work summarizes features of temporalized facts from both diachronic and synchronic perspectives: (1) Diachronicity. Facts often exhibit varying characteristics and trends across different temporal domains; (2) Synchronicity. In specific temporal contexts, various relations between entities influence each other, generating latent semantics. To track above issues, we design a quaternionbased model, TeDS, which divides timestamps into diachronic and synchronic timestamps to support dual temporal perception: (a) Two composite quaternions fusing time and relation information are generated by reorganizing synchronic timestamp and relation quaternions, and Hamilton operator achieves their interaction. (b) Each time point is sequentially mapped to an angle and converted to scalar component of a quaternion using trigonometric functions to build diachronic timestamps. We then rotate relation by using Hamilton operator between it and diachronic timestamp. In this way, TeDS achieves deep integration of relations and time while accommodating different perspectives. Empirically, TeDS significantly outperforms SOTA models on six benchmarks.

Paperid:544

Authors:Gaurav Menghani · Ravi Kumar · Sanjiv Kumar

Title: LAuReL: Learned Augmented Residual Layer

Abstract:
One of the core pillars of efficient deep learning methods is architectural improvements, such as residual/skip connections, which have led to significantly better model convergence and quality. Since their introduction, residual connections have become ubiquitous not only in convolutional neural networks but also in transformerbased architectures, the backbone of LLMs.In this paper, we introduce the Learned Augmented Residual Layer (LAuReL)---a novel generalization of the canonical residual connection---designed to serve as an in-situ replacement while outperforming it in both model quality and footprint metrics. Our experiments show that LAuReL can enhance model quality for both vision and language models while adding fewer parameters and incurring less latency and memory overhead than naively increasing parameter count.For example, on the ImageNet-1K task, LAuReL achieves the same model quality improvements as naively adding an extra layer while using 2.6x fewer parameters. Similarly, when pre-training 1B and 4B parameter LLMs, LAuReL improves performance on a variety of challenging downstream evaluation tasks by 2.54% to 20.05%, while adding only 0.012% and 0.1% additional parameters, respectively.

Paperid:545

Authors:Yejiang Wang · Yuhai Zhao · Zhengkui Wang · Wen Shan · Ling Li · Qian Li · Miaomiao Huang · Meixia Wang · Shirui Pan · Xingwei Wang

Title: N2GON: Neural Networks for Graph-of-Net with Position Awareness

Abstract:
Graphs, fundamental in modeling various research subjects, consist of nodes linked by edges. However, they typically function as components within larger structures in realworld scenarios, such as in protein-protein interactions where each protein is a graph in a larger network. This study delves into the Graph-of-Net (GON), a structure that extends the concept of traditional graphs by representing each node as a graph itself. It provides a multi-level perspective on the relationships between objects, encapsulating both the detailed structure of individual nodes and the broader network of dependencies. To learn node representations within the GON, we propose an position-aware neural network for Graph-of-Net which processes both intra-graph and inter-graph connections and incorporates additional data like node labels. Our model employs dual encoders and graph constructors to build and refine a constraint network, where nodes are adaptively arranged based on their positions, as determined by the network's constraint system. Our model demonstrates significant improvements over baselines in empirical evaluations on various datasets.

Paperid:546

Authors:Ruiyuan Huang · Zengfeng Huang

Title: High Probability Bound for Cross-Learning Contextual Bandits with Unknown Context Distributions

Abstract:
Motivated by applications in online bidding and sleeping bandits, we examine the problem of contextual bandits with cross learning, where the learner observes the loss associated with the action across all possible contexts, not just the current round’s context. Our focus is on a setting where losses are chosen adversarially, and contexts are sampled i.i.d. from a specific distribution. This problem was first studied by Balseiro et al. (2019), who proposed an algorithm that achieves nearoptimal regret under the assumption that the context distribution is known in advance. However, this assumption is often unrealistic. To address this issue, Schneider & Zimmert (2023) recently proposed a new algorithm that achieves nearly optimal expected regret. It is well-known that expected regret can be significantly weaker than high-probability bounds. In this paper, we present a novel, in-depth analysis of their algorithm and demonstrate that it actually achieves near-optimal regret with high probability. There are steps in the original analysis by Schneider & Zimmert (2023) that lead only to an expected bound by nature. In our analysis, we introduce several new insights. Specifically, we make extensive use of the weak dependency structure between different epochs, which was overlooked in previous analyses. Additionally, standard martingale inequalities are not directly applicable, so we refine martingale inequalities to complete our analysis.

Paperid:547

Authors:Kian Kenyon-Dean · Zitong Jerry Wang · John Urbanik · Konstantin Donhauser · Jason Hartford · Saber Saberian · Nil Sahin · Ihab Bendidi · Safiye Celik · Juan Vera · Marta Fay · Imran Haque · Oren Kraus

Title: ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy

Abstract:
Deriving insights from experimentally generated datasets requires methods that can account for random and systematic measurement errors and remove them in order to accurately represent the underlying effects of the conditions being tested. Here we present a framework for pretraining on largescale microscopy datasets that includes three steps: (1) curating a set of diverse and self-consistent training samples, (2) scaling training of an appropriate foundation model architecture on this dataset, (3) evaluating intermediate layers of the trained model to identify the best representation for downstream tasks. Using this strategy, we present the largest foundation model for cell microscopy data to our knowledge, a new 1.9 billion-parameter ViT-G/8 MAE trained on over 8 billion microscopy image crops. Compared to a previous published ViT-L/8 MAE, our new model achieves a 60% improvement in linear separability of genetic perturbations and obtains the best overall performance on whole-genome relationship recall, batch correction replicate consistency, and compound-gene activity prediction benchmarks.

Paperid:548

Authors:Jae-Hong Lee

Title: Bayesian Weight Enhancement with Stead-State Adaptation for Test-time Adaptation in Dynamic Environments

Abstract:
Testtime adaptation (TTA) addresses the machine learning challenge of adapting models to unlabeled test data from shifting distributions in dynamic environments. A key issue in this online setting arises from using unsupervised learning techniques, which introduce explicit gradient noise that degrades model weights. To invest in weight degradation, we propose a Bayesian weight enhancement framework, which generalizes existing weight-based TTA methods that effectively mitigate the issue. Our framework enables robust adaptation to distribution shifts by accounting for diverse weights by modeling weight distributions.Building on our framework, we identify a key limitation in existing methods: their neglect of time-varying covariance reflects the influence of the gradient noise. To address this gap, we propose a novel steady-state adaptation (SSA) algorithm that balances covariance dynamics during adaptation. SSA is derived through the solution of a stochastic differential equation for the TTA process and online inference. The resulting algorithm incorporates a covariance-aware learning rate adjustment mechanism. Through extensive experiments, we demonstrate that SSA consistently improves state-of-the-art methods in various TTA scenarios, datasets, and model architectures, establishing its effectiveness in instability and adaptability.

Paperid:549

Authors:Boyang Sun · Yu Yao · Xinshuai Dong · Zongfang Liu · Tongliang Liu · Yumou Qiu · Kun Zhang

Title: A Sample Efficient Conditional Independence Test in the Presence of Discretization

Abstract:
Conditional independence (CI) test is a fundamental concept in statistics. In many realworld scenarios, some variables may be difficult to measure accurately, often leading to data being represented as discretized values. Applying CI tests directly to discretized data, however, can lead to incorrect conclusions about the independence of latent variables. To address this, recent advancements have sought to infer the correct CI relationship between the latent variables by binarizing the observed data. However, this process results in a loss of information, which degrades the test's performance, particularly with small sample sizes. Motivated by this, this paper introduces a new sample-efficient CI test that does not rely on the binarization process. We find that the relationship can be established by addressing an \textit{over-identifying} restriction problem with \textit{Generalized Method of Moments} (GMM). Based on this finding, we have designed a new test statistic, and its asymptotic distribution has been derived. Empirical results across various datasets show that our method consistently outperforms existing ones.

Paperid:550

Authors:Subhash Kantamneni · Josh Engels · Senthooran Rajamanoharan · Max Tegmark · Neel Nanda

Title: Are Sparse Autoencoders Useful? A Case Study in Sparse Probing

Abstract:
Sparse autoencoders (SAEs) are a popular method for interpreting concepts represented in large language model (LLM) activations. However, there is a lack of evidence regarding the validity of their interpretations due to the lack of a ground truth for the concepts used by an LLM, and a growing number of works have presented problems with current SAEs. One alternative source of evidence would be demonstrating that SAEs improve performance on downstream tasks beyond existing baselines. We test this by applying SAEs to the realworld task of LLM activation probing in four regimes: data scarcity, class imbalance, label noise, and covariate shift. Due to the difficulty of detecting concepts in these challenging settings, we hypothesize that SAEs’ basis of interpretable, concept-level latents should provide a useful inductive bias. However, although SAEs occasionally perform better than baselines on individual datasets, we are unable to design ensemble methods combining SAEs with baselines that consistently outperform ensemble methods solely using baselines. Additionally, although SAEs initially appear promising for identifying spurious correlations, detecting poor dataset quality, and training multi-token probes, we are able to achieve similar results with simple non-SAE baselines as well. Though we cannot discount SAEs’ utility on other tasks, our findings highlight the shortcomings of current SAEs and the need to rigorously evaluate interpretability methods on downstream tasks with strong baselines.

Paperid:551

Authors:Guixiang Wang · Jianjun Li

Title: Large Displacement Motion Transfer with Unsupervised Anytime Interpolation

Abstract:
Motion transfer is to transfer pose in driving video to object of source image, so that object of source image moves. Although great progress has been made recently in unsupervised motion transfer, many unsupervised methods still struggle to accurately model large displacement motions when large motion differences occur between source and driving images. To solve the problem, we propose an unsupervised anytime interpolation based large displacement motion transfer method, which can generate a series of anytime interpolated images between source and driving images. By decomposing large displacement motion into many small displacement motions, difficulty of large displacement motion estimation is reduced. In the process, we design a selector that can select optimal interpolated image from generated interpolated images for downstream tasks. Since there are no real images as labels in the interpolation process, we propose a bidirectional training strategy. Some constraints are added to optimal interpolated image to generate a reasonable interpolated image. To encourage network to generate highquality images, a pre-trained Vision Transformer model is used to design constraint losses. Finally, experiments show that compared with the large displacement motion between source and driving images, small displacement motion between interpolated and driving images is easier to realize motion transfer. Compared with existing state-of-art methods, our method has significant improvements in motion-related metrics.

Paperid:552

Authors:Javan Tahir · Surya Ganguli · Grant Rotskoff

Title: Features are fate: a theory of transfer learning in high-dimensional regression

Abstract:
With the emergence of largescale pre-trained neural networks, methods to adapt such foundation models to data-limited downstream tasks have become a necessity. Fine-tuning, preference optimization, and transfer learning have all been successfully employed for these purposes when the target task closely resembles the source task, but a precise theoretical understanding of task similarity is still lacking. We adopt afeature-centricviewpoint on transfer learning and establish a number of theoretical results that demonstrate that when the target task is well represented by the feature space of the pre-trained model, transfer learning outperforms training from scratch. We study deep linear networks as a minimal model of transfer learning in which we can analytically characterize the transferability phase diagram as a function of the target dataset size and the feature space overlap. For this model, we establish rigorously that when the feature space overlap between the source and target tasks is sufficiently strong, both linear transfer and fine-tuning improve performance, especially in the low data limit. These results build on an emerging understanding of feature learning dynamics in deep linear networks, and we demonstrate numerically that the rigorous results we derive for the linear case also apply to nonlinear networks.

Paperid:553

Authors:David Reber · Sean Richardson · Todd Nief · Cristina Garbacea · Victor Veitch

Title: RATE: Causal Explainability of Reward Models with Imperfect Counterfactuals

Abstract:
Reward models are widely used as proxies for human preferences when aligning or evaluating LLMs.However, reward models are black boxes, and it is often unclear what, exactly, they are actually rewarding. In this paper we develop Rewritebased Attribute Treatment Estimator (RATE) as an effective method for measuring the sensitivity of a reward model to high-level attributes of responses, such as sentiment, helpfulness, or complexity. Importantly, RATE measures thecausaleffect of an attribute on the reward. RATE uses LLMs to rewrite responses to produce imperfect counterfactuals examples that can be used to measure causal effects. A key challenge is that these rewrites are imperfect in a manner that can induce substantial bias in the estimated sensitivity of the reward model to the attribute. The core idea of RATE is to adjust for this imperfect-rewrite effect by rewritingtwice. We establish the validity of the RATE procedure and show empirically that it is an effective estimator.

Paperid:554

Authors:Yung-Sung Chuang · Benjamin Cohen-Wang · Shannon Shen · Zhaofeng Wu · Hu Xu · Xi Victoria Lin · Shang-Wen Li · James Glass · Scott Yih

Title: SelfCite: Self-Supervised Alignment for Context Attribution in Large Language Models

Abstract:
We introduce SelfCite, a novel selfsupervised approach that aligns LLMs to generate high-quality, fine-grained, sentence-level citations for the statements in their generated responses. Instead of only relying on costly and labor-intensive annotations, SelfCite leverages a reward signal provided by the LLM itself throughcontext ablation: If a citation is necessary, removing the cited text from the context should prevent the same response; if sufficient, retaining the cited text alone should preserve the same response. This reward can guide the inference-time best-of-N sampling strategy to improve citation quality significantly, as well as be used in preference optimization to directly fine-tune the models for generating better citations. The effectiveness of SelfCite is demonstrated by increasing citation F1 up to 5.3 points on the LongBench-Cite benchmark across five long-form question answering tasks.

Paperid:555

Authors:Zhitong Xu · Da Long · Yiming Xu · Guang Yang · Shandian Zhe · Houman Owhadi

Title: Toward Efficient Kernel-Based Solvers for Nonlinear PDEs

Abstract:
We introduce a novel kernel learning framework toward efficiently solving nonlinear partial differential equations (PDEs). In contrast to the stateof-the-art kernel solver that embeds differential operators within kernels, posing challenges with a large number of collocation points, our approach eliminates these operators from the kernel. We model the solution using a standard kernel interpolation form and differentiate the interpolant to compute the derivatives. Our framework obviates the need for complex Gram matrix construction between solutions and their derivatives, allowing for a straightforward implementation and scalable computation. As an instance, we allocate the collocation points on a grid and adopt a product kernel, which yields a Kronecker product structure in the interpolation. This structure enables us to avoid computing the full Gram matrix, reducing costs and scaling efficiently to a large number of collocation points. We provide a proof of the convergence and rate analysis of our method under appropriate regularity assumptions. In numerical experiments, we demonstrate the advantages of our method in solving several benchmark PDEs.

Paperid:556

Authors:Khai Nguyen · Hai Nguyen · Tuan Pham · Nhat Ho

Title: Lightspeed Geometric Dataset Distance via Sliced Optimal Transport

Abstract:
We introduce sliced optimal transport dataset distance (sOTDD), a model-agnostic, embedding-agnostic approach for dataset comparison that requires no training, is robust to variations in the number of classes, and can handle disjoint label sets. The core innovation is Moment Transform Projection (MTP), which maps a label, represented as a distribution over features, to a real number. Using MTP, we derive a data point projection that transforms datasets into one-dimensional distributions. The s-OTDD is defined as the expected Wasserstein distance between the projected distributions, with respect to random projection parameters. Leveraging the closed form solution of one-dimensional optimal transport, s-OTDD achieves (near-)linear computational complexity in the number of data points and feature dimensions and is independent of the number of classes. With its geometrically meaningful projection, s-OTDD strongly correlates with the optimal transport dataset distance while being more efficient than existing dataset discrepancy measures. Moreover, it correlates well with the performance gap in transfer learning and classification accuracy in data augmentation.

Paperid:557

Authors:Jiawei Ge · Yuanhao Wang · Wenzhe Li · Chi Jin

Title: Securing Equal Share: A Principled Approach for Learning Multiplayer Symmetric Games

Abstract:
Abstract:This paper examines multiplayer symmetric constantsum games with more than two players in a competitive setting, such as Mahjong, Poker, and various board and video games. In contrast to two-player zero-sum games, equilibria in multiplayer games are neither unique nor non-exploitable, failing to provide meaningful guarantees when competing against opponents who play different equilibria or non-equilibrium strategies. This gives rise to a series of long-lasting fundamental questions in multiplayer games regarding suitable objectives, solution concepts, and principled algorithms. This paper takes an initial step towards addressing these challenges by focusing on the natural objective of *equal share*—securing an expected payoff of $C/n$ in an $n$-player symmetric game with a total payoff of $C$. We rigorously identify the theoretical conditions under which achieving an equal share is tractable and design a series of efficient algorithms, inspired by no-regret learning, that *provably* attain approximate equal share across various settings. Furthermore, we provide complementary lower bounds that justify the sharpness of our theoretical results. Our experimental results highlight worst-case scenarios where meta-algorithms from prior state-of-the-art systems for multiplayer games fail to secure an equal share, while our algorithm succeeds, demonstrating the effectiveness of our approach.

Paperid:558

Authors:Supratim Shit · Gurmehak chadha · Surendra kumar · Bapi Chatterjee

Title: Improved Coresets for Vertical Federated Learning: Regularized Linear and Logistic Regressions

Abstract:
Coreset, as a summary of training data, offers an efficient approach for reducing data processing and storage complexity during training. In the emerging vertical federated learning (VFL) setting, where scattered clients store different data features, it directly reduces communication complexity. In this work, we introduce coresets construction for regularized logistic regression both in centralized and VFL settings. Additionally, we improve the coreset size for regularized linear regression in the VFL setting. We also eliminate the dependency of the coreset size on a property of the data due to the VFL setting. The improvement in the coreset sizes is due to our novel coreset construction algorithms that capture the reduced model complexity due to the added regularization and its subsequent analysis. In experiments, we provide extensive empirical evaluation that backs our theoretical claims. We also compare the performance of our coresets in terms of the distance between the models trained on the complete data and coreset.

Paperid:559

Authors:Raja Sunkara · Ardhendu Tripathy

Title: Adaptive Partitioning Schemes for Optimistic Optimization

Abstract:
Abstract:Applications such as engineering design often require us to optimize a blackbox function, i.e., a system whose inner processing is not analytically known and whose gradients are not available. Practitioners often have a fixed budget for the number of function evaluations and the performance of an optimization algorithm is measured by its simple regret. In this paper, we study the class of ``Optimistic Optimization'' algorithms for black-box optimization that use a partitioning scheme for the domain. We develop algorithms that learn a good partitioning scheme and use flexible surrogate models such as neural networks in the optimization procedure. For multi-index functions on an $m$-dimensional subspace within $d$ dimensions, our algorithm attains $\tilde{O}(n^{-\beta / d})$ regret, where $\beta = 1 + \frac{d-m}{2m-1}$, as opposed to $\tilde{O}(n^{-1/d})$ for SequOOL, a state-of-the-art optimistic optimization algorithm. Our approach is competitive across a wide range of numerical benchmarks. Additionally, we introduce weight quantization in a large language model as a novel task for black-box optimization. Our approach improves the quality of Activation-aware Weight Quantization (AWQ) of the OPT-1.3B model, achieving an approximate 10\% improvement in performance relative to the best possible unquantized model.

Paperid:560

Authors:Nate Veldt · Thomas Stanley · Benjamin Priest · Trevor Steil · Keita Iwabuchi · T.S. Jayram · Geoffrey Sanders

Title: Approximate Forest Completion and Learning-Augmented Algorithms for Metric Minimum Spanning Trees

Abstract:
Abstract:Finding a minimum spanning tree (MST) for $n$ points in an arbitrary metric space is a fundamental primitive for hierarchical clustering and many other ML tasks, but this takes $\Omega(n^2)$ time to even approximate. We introduce a framework for metric MSTs that first (1) finds a forest of trees using practical heuristics, and then (2) finds a small weight set of edges to connect disjoint components in the forest into a spanning tree. We prove that optimally solving step (2) still takes $\Omega(n^2)$ time, but we provide a subquadratic 2.62approximation algorithm. In the spirit of learning-augmented algorithms, we then show that if the heuristic forest found in step (1) overlaps with an optimal MST, we can approximate the original MST problem in subquadratic time, where the approximation factor depends on a measure of overlap. In practice, we find nearly optimal spanning trees for a wide range of metrics, while being orders of magnitude faster than exact algorithms.

Paperid:561

Authors:Charita Dellaporta · Patrick O'Hara · Theodoros Damoulas

Title: Decision Making under the Exponential Family: Distributionally Robust Optimisation with Bayesian Ambiguity Sets

Abstract:
Decision making under uncertainty is challenging as the datagenerating process (DGP) is often unknown. Bayesian inference proceeds by estimating the DGP through posterior beliefs on the model’s parameters. However, minimising the expected risk under these beliefs can lead to suboptimal decisions due to model uncertainty or limited, noisy observations. To address this, we introduce Distributionally Robust Optimisation with Bayesian Ambiguity Sets (DRO-BAS) which hedges against model uncertainty by optimising the worst-case risk over a posterior-informed ambiguity set. We provide two such sets, based on the posterior expectation (DRO-BAS(PE)) or the posterior predictive (DRO-BAS(PP)) and prove that both admit, under conditions, strong dual formulations leading to efficient single-stage stochastic programs which are solved with a sample average approximation. For DRO-BAS(PE), this covers all conjugate exponential family members while for DRO-BAS(PP) this is shown under conditions on the predictive's moment generating function. Our DRO-BAS formulations outperform existing Bayesian DRO on the Newsvendor problem and achieve faster solve times with comparable robustness on the Portfolio problem.

Paperid:562

Authors:Weiqiu You · Helen Qu · Marco Gatti · Bhuvnesh Jain · Eric Wong

Title: Sum-of-Parts: Self-Attributing Neural Networks with End-to-End Learning of Feature Groups

Abstract:
Selfattributing neural networks (SANNs) present a potential path towards interpretable models for high-dimensional problems, but often face significant trade-offs in performance. In this work, we formally prove a lower bound on errors of per-feature SANNs, whereas group-based SANNs can achieve zero error and thus high performance. Motivated by these insights, we propose Sum-of-Parts (SOP), a framework that transforms any differentiable model into a group-based SANN, where feature groups are learned end-to-end without group supervision. SOP achieves state-of-the-art performance for SANNs on vision and language tasks, and we validate that the groups are interpretable on a range of quantitative and semantic metrics. We further validate the utility of SOP explanations in model debugging and cosmological scientific discovery.

Paperid:563

Authors:Liam Hodgkinson · Zhichao Wang · Michael Mahoney

Title: Models of Heavy-Tailed Mechanistic Universality

Abstract:
Recent theoretical and empirical successes in deep learning, including the celebrated neural scaling laws, are punctuated by observations that many objects of interest tend to exhibit some form of heavytailed or power law behavior. In particular, the prevalence of heavy-tailed spectral densities in the Jacobian, Hessian, and weight matrices of deep neural networks has recently been coined asheavy-tailed mechanistic universality(HT-MU). Empirical evidence suggests a robust correlation between power law exponents and model performance, indicating that HT-MU may be a fundamental aspect of deep learning efficacy. Here, we propose a general family of random feature matrix models---thehigh-temperature inverse-Wishart ensemble---to explore the attributes which give rise to heavy-tailed behavior in trained neural network models. Under this model, spectral densities with power laws on both tails arise through a combination of three independent factors: (i) complex correlation structures in the data; (ii) reduced temperatures during training; and (iii) reduced eigenvector entropy appearing as implicit bias in the model structure. Implications of our model on other appearances of heavy tails, including the five phases of learning, neural scaling laws, and optimizer trajectories, are also discussed.

Paperid:564

Authors:Dihan Zheng · Bo Huang

Title: Bridging Protein Sequences and Microscopy Images with Unified Diffusion Models

Abstract:
Fluorescence microscopy is ubiquitously used in cell biology research to characterize the cellular role of a protein. To help elucidate the relationship between the amino acid sequence of a protein and its cellular function, we introduce CELLDiff, a unified diffusion model facilitating bidirectional transformations between protein sequences and their corresponding microscopy images. Utilizing reference cell morphology images and a protein sequence, CELL-Diff efficiently generates corresponding protein images. Conversely, given a protein image, the model outputs protein sequences. CELL-Diff integrates continuous and diffusion models within a unified framework and is implemented using a transformer-based network. We train CELL-Diff on the Human Protein Atlas (HPA) dataset and fine-tune it on the OpenCell dataset. Experimental results demonstrate that CELL-Diff outperforms existing methods in generating high-fidelity protein images, making it a practical tool for investigating subcellular protein localization and interactions.

Paperid:565

Authors:Mengmeng Ma · Tang Li · Yunxiang Peng · LIN LU · Volkan Beylergil · Binsheng Zhao · Oguz Akin · Xi Peng

Title: "Why There Is a Tumor?": Safeguard Tumor Segmentation and Detection with Trustworthy Rationales

Abstract:
Medical AI models excel at tumor detection and segmentation. However, their latent representations often lack explicit ties to clinical semantics, producing outputs less trusted in clinical practice. Most of the existing models generate either segmentation masks/labels (localizing where without why) or textual justifications (explaining why without where), failing to ground clinical concepts in spatially localized evidence. To bridge this gap, we propose to develop models that can justify the segmentation or detection using clinically relevant terms and point to visual evidence. We address two core challenges: First, we curate a rationale dataset to tackle the lack of paired images, annotations, and textual rationales for training. The dataset includes 180K imagemask-rationale triples with quality evaluated by expert radiologists. Second, we design rationale-informed optimization that disentangles and localizes fine-grained clinical concepts in a self-supervised manner without requiring pixel-level concept annotations. Experiments across medical benchmarks show our model demonstrates superior performance in segmentation, detection, and beyond. The anonymous link to our code.

Paperid:566

Authors:Hang Zhou · Yuezhou Ma · Haixu Wu · Haowen Wang · Mingsheng Long

Title: Unisolver: PDE-Conditional Transformers Are Universal Neural PDE Solvers

Abstract:
Deep models have recently emerged as promising tools to solve partial differential equations (PDEs), known as neural PDE solvers. While neural solvers trained from either simulation data or physicsinformed loss can solve PDEs reasonably well, they are mainly restricted to a few instances of PDEs, e.g. a certain equation with a limited set of coefficients. This limits their generalization to diverse PDEs, preventing them from being practical surrogate models for classical numerical solvers. In this paper, we present the Universal Neural PDE Solver (Unisolver) capable of solving a wide scope of PDEs by training a novel Transformer model on diverse data and conditioned on diverse PDEs. Instead of purely scaling up data and parameters, Unisolver stems from the theoretical analysis of the PDE-solving process. Inspired by the mathematical structure of PDEs that a PDE solution is fundamentally governed by a series of PDE components such as equation symbols, coefficients, and boundary conditions, we define a complete set of PDE components and flexibly embed them as domain-wise and point-wise deep conditions for Transformer PDE solvers. Integrating physical insights with recent Transformer advances, Unisolver achieves consistent state-of-the-art results on three challenging large-scale benchmarks, showing impressive performance gains and favorable PDE generalizability.

Paperid:567

Authors:Sai Advaith Maddipatla · Nadav Bojan · Meital Bojan · Sanketh Vedula · Paul Schanda · Ailie Marx · Alexander Bronstein

Title: Inverse problems with experiment-guided AlphaFold

Abstract:
Proteins exist as a dynamic ensemble of multiple conformations, and these motions are often crucial for their functions. However, current structure prediction methods predominantly yield a single conformation, overlooking the conformational heterogeneity revealed by diverse experimental modalities. Here, we present a framework for building experimentgrounded protein structure generative models that infer conformational ensembles consistent with measured experimental data. The key idea is to treat state-of-the-art protein structure predictors (e.g., AlphaFold3) as sequence-conditioned structural priors, and cast ensemble modeling as posterior inference of protein structures given experimental measurements. Through extensive real-data experiments, we demonstrate the generality of our method to incorporate a variety of experimental measurements. In particular, our framework uncovers previously unmodeled conformational heterogeneity from crystallographic densities, generates high-accuracy NMR ensembles orders of magnitude faster than status quo, and incorporates pairwise cross-link constraints. Notably, we demonstrate that our ensembles outperform AlphaFold3 and sometimes better fit experimental data than publicly deposited structures to the protein database (PDB). We believe that this approach will unlock building predictive models that fully embrace experimentally observed conformational diversity.

Paperid:568

Authors:Fangxu Yu · Lai Jiang · Haoqiang Kang · Shibo Hao · Lianhui Qin

Title: Flow of Reasoning: Training LLMs for Divergent Reasoning with Minimal Examples

Abstract:
The ability to generate diverse solutions to a given problem is a hallmark of human creativity. This divergent reasoning is also crucial for machines, enhancing their robustness and enabling them to assist humans in many applications such as scientific discovery. However, existing approaches to multistep reasoning with large language models (LLMs) have mostly focused only on reasoning accuracy, without further discovering more diverse valid solutions. For example, supervised fine-tuning can improve LLM reasoning quality, but requires extensive supervised data to capture the full range of possible solutions. Reinforcement learning aims to find limited highest-reward solutions while neglecting the solution diversity. To fill this gap, we propose Flow of Reasoning (FoR), an efficient diversity-seeking LLM finetuning method aimed at improving reasoning quality and diversity with minimal data. FoR formulates multi-step LLM reasoning as a Markovian flow on a DAG-structured reasoning graph. This formulation allows us to incorporate and adapt principled GFlowNet approaches, for finetuning LLMs to sample divergent paths with probabilities proportional to the (unnormalized) reward of target problems. Extensive experiments show that, with limited training examples (e.g., 15 examples), FoR enables the discovery of diverse, creative, high-quality solutions, greatly outperforming a wide range of existing inference and training methods across six challenging reasoning tasks, including BlocksWorld (embodied reasoning), Game24 (math puzzle solving), Rubik's Cube (spatial reasoning), 1D-ARC (abstraction reasoning), GSM8k (math reasoning), and ProntoQA (logical reasoning).

Paperid:569

Authors:Daniil Tiapkin · Daniele Calandriello · Johan Ferret · Sarah Perrin · Nino Vieillard · Alexandre Rame · Mathieu Blondel

Title: On Teacher Hacking in Language Model Distillation

Abstract:
Posttraining of language models (LMs) increasingly relies on the following two stages: (i) knowledge distillation, where the LM is trained to imitate a larger teacher LM, and (ii) reinforcement learning from human feedback (RLHF), where the LM is aligned by optimizing a reward model. In the second RLHF stage, a well-known challenge is reward hacking, where the LM over-optimizes the reward model, leading to degraded performance on the true objective, in line with Goodhart's law.In this paper, we investigate whether a similar phenomenon, that we call teacher hacking, can occur during knowledge distillation. This could arise because the teacher LM is itself an imperfect approximation of the true distribution. To study this, we propose a controlled experimental setup involving: (i) an oracle LM representing the ground-truth distribution, (ii) a teacher LM distilled from the oracle, and (iii) a student LM distilled from the teacher.Our experiments reveal the following insights. When using a fixed offline dataset for distillation, teacher hacking occurs; moreover, we can detect it by observing when the optimization process deviates from polynomial convergence laws. In contrast, employing online data generation techniques effectively mitigates teacher hacking. More precisely, we identify data diversity as the key factor in preventing hacking.Overall, our findings provide a deeper understanding of the benefits and limitations of distillation for building robust LMs.

Paperid:570

Authors:chengqian gao · Haonan Li · Liu Liu · Zeke Xie · Peilin Zhao · Zhiqiang Xu

Title: Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Abstract:
The alignment of large language models (LLMs) often assumes that using more clean data yields better outcomes, overlooking the match between model capacity and example difficulty. Challenging this, we propose a new principle:Preference data vary in difficulty, and overly difficult examples hinder alignment, by exceeding the model's capacity. Through systematic experimentation, we validate this principle with three key findings: (1) preference examples vary in difficulty, as evidenced by consistent learning orders across alignment runs; (2) overly difficult examples significantly degrade performance across four LLMs and two datasets; and (3) the capacity of a model dictates its threshold for handling difficult examples, underscoring a critical relationship between data selection and model capacity. Building on this principle, we introduceSelective DPO, which filters out overly difficult examples. This simple adjustment improves alignment performance by 916\% in win rates on the AlpacaEval 2 benchmark compared to the DPO baseline, suppressing a series of DPO variants with different algorithmic adjustments. Together, these results illuminate the importance of aligning data difficulty with model capacity, offering a transformative perspective for improving alignment strategies in LLMs.

Paperid:571

Authors:Huanjian Zhou · Masashi Sugiyama

Title: Parallel Simulation for Sampling under Isoperimetry and Score-based Diffusion Models

Abstract:
Abstract:Sampling from highdimensional probability distributions is fundamental in machine learning and statistics. As datasets grow larger, computational efficiency becomes increasingly important, particularly in reducing \emph{adaptive complexity}, namely the number of sequential rounds required for sampling algorithms. While recent works have introduced several parallelizable techniques, they often exhibit suboptimal convergence rates and remain significantly weaker than the latest lower bounds for log-concave sampling.To address this, we propose a novel parallel sampling method that improves adaptive complexity dependence on dimension $d$ reducing it from $\widetilde{\mathcal{O}}(\log^2 d)$ to $\widetilde{\mathcal{O}}(\log d)$, which is even optimal for log-concave sampling with some specific adaptive complexity. Our approach builds on parallel simulation techniques from scientific computing.

Paperid:572

Authors:Wenhui Zhu · Peijie Qiu · Xiwen Chen · Zhangsihao Yang · Aristeidis Sotiras · Abolfazl Razi · Yalin Wang

Title: How Effective Can Dropout Be in Multiple Instance Learning ?