ICML

Beyond Self-Repellent Kernels: History-Driven Target Towards Efficient Nonlinear MCMC on General Graphs

Jie Hu, Yi-Ting Ma, Do Young Eun

机构 * Department of Electrical and Computer Engineering, North Carolina State University, Raleigh, North Carolina, USA（电气与计算机工程系，北卡罗来纳州立大学，拉法叶，北卡罗来纳州，美国） 纠错

Comments Accepted at ICML 2025 (Oral)

Preference learning made easy: Everything should be understood through win rate

Lily H. Zhang, Rajesh Ranganath

机构 * Center for Data Science, New York University, New York, USA（数据科学中心，纽约大学，纽约，美国） ； Courant Institute, New York University, New York, USA（柯朗研究所，纽约大学，纽约，美国） 纠错

Comments ICML 2025

NestQuant: Nested Lattice Quantization for Matrix Products and LLMs

Semyon Savkin, Eitan Porat, Or Ordentlich, Yury Polyanskiy

机构 * MIT（麻省理工学院） ； Hebrew University of Jerusalem（耶路撒冷希伯来大学） 纠错

Comments 23 pages; Accepted at the 42nd International Conference on Machine Learning (ICML 2025)

Compressed Image Generation with Denoising Diffusion Codebook Models

Guy Ohayon, Hila Manor, Tomer Michaeli, Michael Elad

机构 * Faculty of Computer Science, Technion -- Israel Institute of Technology, Haifa, Israel（计算机科学系，技术离子理工学院——以色列理工学院，海法，以色列） ； Faculty of Electrical and Computer Engineering, Technion -- Israel Institute of Technology, Haifa, Israel（电气与计算机工程系，技术离子理工学院——以色列理工学院，海法，以色列） 纠错

Comments Published in the International Conference on Machine Learning (ICML) 2025. Code and demo are available at https://ddcm-2025.github.io/

Geometric Representation Condition Improves Equivariant Molecule Generation

Zian Li, Cai Zhou, Xiyuan Wang, Xingang Peng, Muhan Zhang

机构 * Institute for Artificial Intelligence, Peking University, Beijing, China（北京大学人工智能研究院） ； School of Intelligence Science and Technology, Peking University, Beijing, China（北京大学智能科学与技术学院） ； Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, USA（麻省理工学院电子工程与计算机科学系） ； Department of Automation, Tsinghua University, Beijing, China（清华大学自动化系） 纠错

Comments Accepted to ICML 2025 as a Spotlight Poster

Selective Prompt Anchoring for Code Generation

Yuan Tian, Tianyi Zhang

机构 * Purdue University（普渡大学） 纠错

Comments Accepted by ICML'25

Density Ratio Estimation-based Bayesian Optimization with Semi-Supervised Learning

Jungtaek Kim

机构 * University of Wisconsin--Madison（威斯康星大学麦迪逊分校） 纠错

Comments Accepted at the 42nd International Conference on Machine Learning (ICML 2025)

Agent WARPP: Workflow Adherence via Runtime Parallel Personalization

Maria Emilia Mazzolenis, Ruirui Zhang

Comments Accepted at the ICML 2025 Workshop on Multi-Agent Systems in the Era of Foundation Models: Opportunities, Challenges, and Futures. Code repo: https://github.com/emiliamazzo/WARPP/

BEAVER: Building Environments with Assessable Variation for Evaluating Multi-Objective Reinforcement Learning

Ruohong Liu, Jack Umenberger, Yize Chen

机构 * University of Oxford（牛津大学） ； University of Alberta（阿尔伯塔大学） 纠错

Comments Accepted at the Workshop on Computational Optimization of Buildings (ICML CO-BUILD), 42nd International Conference on Machine Learning (ICML 2025), Vancouver, Canada

LaMAGIC2: Advanced Circuit Formulations for Language Model-Based Analog Topology Generation

Chen-Chia Chang, Wan-Hsuan Lin, Yikang Shen, Yiran Chen, Xin Zhang

机构 * IBM T. J. Watson Research Center（IBM T.J.沃森研究中心） ； Duke University（杜克大学） ； MIT-IBM Watson AI Lab（麻省理工-IBM沃森人工智能实验室） ； University of California, Los Angeles（加州大学洛杉矶分校） 纠错

Comments Accepted at 42nd International Conference on Machine Learning (ICML) 2025

Journal ref Proceedings of the 42st International Conference on Machine Learning, 2025, PMLR

Video Self-Distillation for Single-Image Encoders: A Step Toward Physically Plausible Perception

Marcel Simon, Tae-Ho Kim, Seul-Ki Yeom

机构 * Nota AI GmbH 纠错

Comments 4 pages, 2 figures, 2 tables

Journal ref 2025 International Conference on Machine Learning Workshop

Fishers for Free? Approximating the Fisher Information Matrix by Recycling the Squared Gradient Accumulator

YuXin Li, Felix Dangel, Derek Tam, Colin Raffel

机构 * University of Toronto \& Vector Institute 纠错

Comments 19 pages, 2 figures. Accepted as a spotlight poster at ICML 2025

Reactivation: Empirical NTK Dynamics Under Task Shifts

Yuzhi Liu, Zixuan Chen, Zirui Zhang, Yufei Liu, Giulia Lanzillotta

机构 * Department of Mathematics, ETH Zurich, Switzerland（苏黎世联邦理工学院数学系） ； Department of Computer Science, ETH Zurich, Switzerland（苏黎世联邦理工学院计算机科学系） 纠错

Comments Accepted by the 3rd Workshop on High-dimensional Learning Dynamics (HiLD), ICML 2025

Scalpel vs. Hammer: GRPO Amplifies Existing Capabilities, SFT Replaces Them

Neel Rajani, Aryo Pradipta Gema, Seraphina Goldfarb-Tarrant, Ivan Titov

机构 * Institute for Language, Cognition and Computation (ILCC), University of Edinburgh, United Kingdom（语言、认知与计算研究所（ILCC），爱丁堡大学，英国） ； Institute for Logic, Language and Computation (ILLC), University of Amsterdam, Netherlands（逻辑、语言与计算研究所（ILLC），阿姆斯特丹大学，荷兰） 纠错

Journal ref Actionable Interpretability Workshop ICML 2025

Exploration Behavior of Untrained Policies

Jacob Adamczyk

机构 * Department of Physics, University of Massachusetts Boston（马萨诸塞大学波士顿分校物理系） 纠错

Comments High-dimensional Learning Dynamics Workshop at ICML-2025

Bridging Quantum and Classical Computing in Drug Design: Architecture Principles for Improved Molecule Generation

Andrew Smith, Erhan Guven

机构 * Whiting School of Engineering, Johns Hopkins University, Baltimore, Maryland（约翰霍普金斯大学工程学院，巴尔的摩，马里兰州） 纠错

Comments Published in Proceedings of the Workshop on Generative AI for Biology at the 42nd International Conference on Machine Learning 10 pages, 7 figures

Journal ref Proceedings of the Workshop on Generative AI for Biology at the 42nd International Conference on Machine Learning. PMLR 267, 2025

Distillation Scaling Laws

Dan Busbridge, Amitis Shidani, Floris Weers, Jason Ramapuram, Etai Littwin, Russ Webb

机构 * Apple（苹果公司） ； University of Oxford, UK（牛津大学） 纠错

Comments Version accepted to ICML 2025. 69 pages, 54 figures, 13 tables

Value-Based Deep RL Scales Predictably

Oleh Rybkin, Michal Nauman, Preston Fu, Charlie Snell, Pieter Abbeel, Sergey Levine, Aviral Kumar

机构 * University of Warsaw（华沙大学） 纠错

Comments ICML 2025

Towards Constraint-aware Learning for Resource Allocation in NFV Networks

Tianfu Wang, Long Yang, Chao Wang, Chuan Qin, Liwei Deng, Wei Wu, Junyang Wang, Li Shen, Hui Xiong

Comments ICML 2025 Workshop on ML4Wireless (Best Paper Award)

Spatial Transformer Network YOLO Model for Agricultural Object Detection

Yash Zambre, Ekdev Rajkitkul, Akshatha Mohan, Joshua Peeples

机构 * 1 Department of Electrical ； Computer Engineering, Texas A\&M University, College Station, TX, USA 2 Department of Computer Science ； Engineering, Texas A\&M University, College Station, TX, USA ； 2 Department of Computer Science 纠错

Comments 7 pages, 5 figures, accepted to 2024 IEEE International Conference on Machine Learning and Applications

Self-Supervised Coarsening of Unstructured Grid with Automatic Differentiation

Sergei Shumilin, Alexander Ryabov, Nikolay Yavich, Evgeny Burnaev, Vladimir Vanovskiy

机构 * Applied AI Center, Skolkovo Institute of Science and Technology, Moscow, Russia（斯克尔科沃科学与技术研究院应用人工智能中心） ； AIRI Institute, Moscow, Russia（莫斯科AIRI研究所） 纠错

Journal ref Proceedings of the 41st International Conference on Machine Learning, PMLR 235:45315-45328, 2024

Are LLM Belief Updates Consistent with Bayes' Theorem?

Sohaib Imran, Ihor Kendiukhov, Matthew Broerman, Aditya Thomas, Riccardo Campanella, Rob Lamb, Peter M. Atkinson

机构 * Lancaster Environment Centre, Lancaster University（兰卡斯特大学环境中心） ； School of Computing and Communications, Lancaster University（兰卡斯特大学计算机与通信学院） ； Environmental Science, University of Southampton（索姆塞特大学环境科学系） ； College of Surveying and Geo-Informatics, Tongji University（同济大学测绘与地理信息学院） ； University of Tuebingen, Department of Computer Science（图宾根大学计算机科学系） ； Graduate School of Natural Sciences, Faculty of Science, Utrecht University（乌得勒支大学自然科学研究生院） 纠错

Comments Accepted at the ICML 2025 Workshop on Assessing World Models

Federated Learning for Large-Scale Cloud Robotic Manipulation: Opportunities and Challenges

Obaidullah Zaland, Chanh Nguyen, Florian T. Pokorny, Monowar Bhuyan

机构 * Department of Computing Science, Umeå University（计算机科学系，乌梅大学） ； Division of Robotics, Perception and Learning, KTH Royal Institute of Technology（机器人、感知与学习 division，皇家理工学院） 纠错

Comments Accepted for Presentation at IEEE International Conference on Machine Learning and Cybernetics (ICMLC) 2025

GenSelect: A Generative Approach to Best-of-N

Shubham Toshniwal, Ivan Sorokin, Aleksander Ficek, Ivan Moshkov, Igor Gitman

机构 * NVIDIA 纠错

Comments Presented at the 2nd AI for MATH Workshop @ ICML

Leveraging the Structure of Medical Data for Improved Representation Learning

Andrea Agostini, Sonia Laguna, Alain Ryser, Samuel Ruiperez-Campillo, Moritz Vandenhirtz, Nicolas Deperrois, Farhad Nooralahzadeh, Michael Krauthammer, Thomas M. Sutter, Julia E. Vogt

机构 * Department of Computer Science, ETH Zurich, Zurich, Switzerland（苏黎世联邦理工学院计算机科学系） ； Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland（苏黎世大学定量生物医学系） 纠错

Journal ref Published at the ICML 2025 Workshop on Multi-modal Foundation Models and Large Language Models for Life Sciences

Unisoma: A Unified Transformer-based Solver for Multi-Solid Systems

Shilong Tao, Zhe Feng, Haonan Sun, Zhanxing Zhu, Yunhuai Liu

机构 * School of Computer Science, Peking University, Beijing, China（计算机科学学院，北京大学，中国） ； School of Electrical and Computer Science, University of Southampton, UK（电子与计算机科学学院，南安普顿大学，英国） 纠错

Comments Proceedings of the 42nd International Conference on Machine Learning

History-Guided Video Diffusion

Kiwhan Song, Boyuan Chen, Max Simchowitz, Yilun Du, Russ Tedrake, Vincent Sitzmann

机构 * MIT（麻省理工学院） ； Harvard University（哈佛大学） ； Carnegie Mellon University（卡内基梅隆大学） 纠错

Comments ICML 2025. Project website: https://boyuan.space/history-guidance

ELITE: Enhanced Language-Image Toxicity Evaluation for Safety

Wonjun Lee, Doehyeon Lee, Eugene Choi, Sangyoon Yu, Ashkan Yousefpour, Haon Park, Bumsub Ham, Suhyun Kim

机构 * Yonsei University（延世大学） ； Kyung Hee University（庆熙大学） ； Korea Institute of Science（韩国科学研究院） ； Seoul National University（首尔国立大学） ； Sookmyung Women's University（_sookmyung女子大学） 纠错

Comments ICML 2025. Project page at https://velpegor.github.io/ELITE/

BlockDialect: Block-wise Fine-grained Mixed Format Quantization for Energy-Efficient LLM Inference

Wonsuk Jang, Thierry Tambe

机构 * Department of Electrical Engineering, Stanford University, CA, USA（电气工程系，斯坦福大学） 纠错

Comments ICML 2025

DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts

Tobias Braun, Mark Rothermel, Marcus Rohrbach, Anna Rohrbach

机构 * Technical University of Darmstadt \& hessian.AI, Germany 纠错

Comments ICML 2025 version. 9 pages main paper, 35 pages with appendix, 18 figures and 7 tables. Corrected two inconsistent numbers in Table 2