ICML

CLARIFY: Contrastive Preference Reinforcement Learning for Untangling Ambiguous Queries

Ni Mu, Hao Hu, Xiao Hu, Yiqin Yang, Bo Xu, Qing-Shan Jia

机构 * Beijing Key Laboratory of Embodied Intelligence Systems, Department of Automation, Tsinghua University, Beijing, China（北京智能体智能系统重点实验室，自动化系，清华大学，北京，中国） ； The Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China（复杂系统认知与决策智能重点实验室，自动化研究所，中国科学院，北京，中国） 纠错

Comments ICML 2025

How Do Images Align and Complement LiDAR? Towards a Harmonized Multi-modal 3D Panoptic Segmentation

Yining Pan, Qiongjie Cui, Xulei Yang, Na Zhao

机构 * Singapore University of Technology and Design (SUTD)（新加坡科技设计大学） ； Institute for Infocomm Research (I2R), A*STAR, Singapore（信息与通信研究院（I2R）） 纠错

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

Beyond Induction Heads: In-Context Meta Learning Induces Multi-Phase Circuit Emergence

Gouki Minegishi, Hiroki Furuta, Shohei Taniguchi, Yusuke Iwasawa, Yutaka Matsuo

机构 * The University of Tokyo（东京大学） 纠错

Comments Accepted to ICML 2025

Compositional Causal Reasoning Evaluation in Language Models

Jacqueline R. M. A. Maasch, Alihan Hüyük, Xinnuo Xu, Aditya V. Nori, Javier Gonzalez

机构 * Harvard University（哈佛大学） ； Microsoft Research Cambridge（微软研究院剑桥） 纠错

Journal ref The 42nd International Conference on Machine Learning (ICML 2025)

Exact Upper and Lower Bounds for the Output Distribution of Neural Networks with Random Inputs

Andrey Kofnov, Daniel Kapla, Ezio Bartocci, Efstathia Bura

机构 * Faculty of Mathematics（数学学院） ； Faculty of Informatics, TU Wien, Vienna, Austria（信息学院，维也纳技术大学，维也纳，奥地利） 纠错

Comments Forthcoming at the 42nd International Conference on Machine Learning (ICML 2025); 9+16 pages

Zero-shot Meta-learning for Tabular Prediction Tasks with Adversarially Pre-trained Transformer

Yulun Wu, Doron L. Bergman

机构 * University of California, Berkeley（加州大学伯克利分校） 纠错

Comments Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025

Calibrated Physics-Informed Uncertainty Quantification

Vignesh Gopakumar, Ander Gray, Lorenzo Zanisi, Timothy Nunn, Daniel Giles, Matt J. Kusner, Stanislas Pamela, Marc Peter Deisenroth

机构 * Centre for Artificial Intelligence, University College London（人工智能研究中心，伦敦大学学院） ； Heudiasyc Laboratory（Heudiasyc实验室） ； Mila - Quebec AI Institute（魁北克人工智能研究所） 纠错

Journal ref ICML 2025

Human-Aligned Image Models Improve Visual Decoding from the Brain

Nona Rajabi, Antônio H. Ribeiro, Miguel Vasco, Farzaneh Taleb, Mårten Björkman, Danica Kragic

机构 * KTH Royal Institute of Technology, Stockholm, Sweden（皇家理工学院） ； Department of Information Technology, Uppsala University, Uppsala, Sweden（信息科技系） 纠错

Comments Accepted to ICML 2025

Federated Generalised Variational Inference: A Robust Probabilistic Federated Learning Framework

Terje Mildner, Oliver Hamelijnck, Paris Giampouras, Theodoros Damoulas

机构 * University of Warwick, Department of Computer Science, Coventry, United Kingdom（沃里克大学计算机科学系） ； University of Warwick, Department of Statistics, Coventry, United Kingdom（沃里克大学统计系） 纠错

Comments Accepted at ICML 2025

Generation from Noisy Examples

Ananth Raman, Vinod Raman

机构 * University of Michigan, Ann Arbor（密歇根大学安娜堡分校） ； Bridgewater-Raritan Regional High School（布里奇沃特-拉里坦地区高中） 纠错

Comments 15 pages. Accepted to ICML 2025

Provably Cost-Sensitive Adversarial Defense via Randomized Smoothing

Yuan Xin, Dingfan Chen, Michael Backes, Xiao Zhang

机构 * CISPA Helmholtz Center for Information Security（信息安全研究中心赫尔姆霍茨中心） ； Max Planck Institute for Intelligent Systems（智能系统马克斯·普朗克研究所） 纠错

Comments Published in ICML 2025

From Debate to Equilibrium: Belief-Driven Multi-Agent LLM Reasoning via Bayesian Nash Equilibrium

Xie Yi, Zhanke Zhou, Chentao Cao, Qiyu Niu, Tongliang Liu, Bo Han

机构 * Academy for Engineering and Technology, Fudan University（复旦大学工程与技术学院） ； TMLR Group, Department of Computer Science, Hong Kong Baptist University（香港 Baptist 大学计算机科学系 TMLR 组） ； Sydney AI Center, The University of Sydney（悉尼大学悉尼人工智能中心） 纠错

Comments Accepted by ICML 2025

Highly Compressed Tokenizer Can Generate Without Training

L. Lao Beyer, T. Li, X. Chen, S. Karaman, K. He

机构 * MIT LIDS（麻省理工学院LIDS） ； MIT CSAIL（麻省理工学院CSAIL） ； Meta FAIR 纠错

Comments Main manuscript: 9 pages, 7 figures. Appendix: 8 pages, 9 figures. To appear in the Proceedings of the 42nd International Conference on Machine Learning

What makes an Ensemble (Un) Interpretable?

Shahaf Bassan, Guy Amir, Meirav Zehavi, Guy Katz

机构 * The Hebrew University of Jerueslaem（耶路撒冷希伯来大学） ； Cornell University（康奈尔大学） ； Ben-Gurion University of the Negev（内盖夫本·古里安大学） 纠错

Comments To appear in ICML 2025

Accelerating Spectral Clustering under Fairness Constraints

Francesco Tonin, Alex Lambert, Johan A. K. Suykens, Volkan Cevher

机构 * EPFL（苏黎世联邦理工学院） 纠错

Comments ICML 2025

BalancEdit: Dynamically Balancing the Generality-Locality Trade-off in Multi-modal Model Editing

Dongliang Guo, Mengxuan Hu, Zihan Guan, Thomas Hartvigsen, Sheng Li

机构 * University of Virginia（弗吉尼亚大学） 纠错

Journal ref Proceedings of the 42 nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025

Self-Supervised Transformers as Iterative Solution Improvers for Constraint Satisfaction

Yudong W. Xu, Wenhao Li, Scott Sanner, Elias B. Khalil

机构 * University of Toronto（多伦多大学） ； Vector Institute for Artificial Intelligence（向量人工智能研究所） ； Scale AI Research Chair in Data-Driven Algorithms for Modern Supply Chains（数据驱动算法现代供应链研究席位） 纠错

Comments ICML 2025

Retrieval-augmented systems can be dangerous medical communicators

Lionel Wong, Ayman Ali, Raymond Xiong, Shannon Zeijang Shen, Yoon Kim, Monica Agrawal

机构 * MIT CSAIL（麻省理工学院计算机科学与人工智能实验室） ； Duke University（杜克大学） ； Stanford University（斯坦福大学） 纠错

Comments Position paper in Proceedings of the 42 nd International Conference on Machine Learning

Distinguishing Cause from Effect with Causal Velocity Models

Johnny Xi, Hugh Dance, Peter Orbanz, Benjamin Bloem-Reddy

机构 * Department of Statistics, University of British Columbia（不列颠哥伦比亚大学统计学系） ； Gatsby Unit, University College London（伦敦大学学院Gatsby单元） 纠错

Comments ICML 2025

From Language Models over Tokens to Language Models over Characters

Tim Vieira, Ben LeBrun, Mario Giulianelli, Juan Luis Gastaldi, Brian DuSell, John Terilla, Timothy J. O'Donnell, Ryan Cotterell

机构 * City University of New York（纽约城市大学） ； McGill University（麦吉尔大学） ； Canada CIFAR AI Chair（加拿大CIFAR人工智能主席） 纠错

Comments ICML 2025

Towards a Mechanistic Explanation of Diffusion Model Generalization

Matthew Niedoba, Berend Zwartsenberg, Kevin Murphy, Frank Wood

机构 * University of British Columbia（不列颠哥伦比亚大学） ； Alberta Machine Intelligence Institute（阿尔伯塔机器智能研究所） 纠错

Comments ICML 2025 Spotlight

MF-LAL: Drug Compound Generation Using Multi-Fidelity Latent Space Active Learning

Peter Eckmann, Dongxia Wu, Germano Heinzelmann, Michael K. Gilson, Rose Yu

机构 * Department of Computer Science and Engineering（计算机科学与工程系） ； Department of Chemistry and Biochemistry（化学与生物化学系） ； Skaggs School of Pharmacy and Pharmaceutical Sciences（斯卡格斯药学与药学科学学校） 纠错

Comments ICML 2025. 9 pages, 5 figures. arXiv admin note: text overlap with arXiv:2402.10387

Positional Attention: Expressivity and Learnability of Algorithmic Computation

Artur Back de Luca, George Giapitzakis, Shenghao Yang, Petar Veličković, Kimon Fountoulakis

机构 * University of Waterloo（滑铁卢大学） ； Google DeepMind（谷歌DeepMind） 纠错

Comments 64 pages, 37 figures, Forty-Second International Conference on Machine Learning (ICML 2025)

Metric Learning for Clifford Group Equivariant Neural Networks

Riccardo Ali, Paulina Kulytė, Haitz Sáez de Ocáriz Borde, Pietro Liò

机构 * University of Cambridge（剑桥大学） ； University of Oxford（牛津大学） 纠错

Comments Workshop on Geometry-grounded Representation Learning and Generative Modeling (GRaM) at the ICML 2024

SK-VQA: Synthetic Knowledge Generation at Scale for Training Context-Augmented Multimodal LLMs

Xin Su, Man Luo, Kris W Pan, Tien Pei Chou, Vasudev Lal, Phillip Howard

机构 * Intel Labs（英特尔实验室） ； Amazon（亚马逊） 纠错

Comments ICML 2025 Spotlight Oral

Correlated Errors in Large Language Models

Elliot Kim, Avi Garg, Kenny Peng, Nikhil Garg

机构 * cornell（康奈尔大学） 纠错

Comments Accepted to ICML 2025

Language Models over Canonical Byte-Pair Encodings

Tim Vieira, Tianyu Liu, Clemente Pasti, Yahya Emara, Brian DuSell, Benjamin LeBrun, Mario Giulianelli, Juan Luis Gastaldi, Timothy J. O'Donnell, Ryan Cotterell

机构 * McGill University（麦吉尔大学） ； Canada CIFAR AI Chair（加拿大 CIFAR 人工智能主席） 纠错

Comments ICML 2025

Statistical Hypothesis Testing for Auditing Robustness in Language Models

Paulius Rauba, Qiyao Wei, Mihaela van der Schaar

机构 * University of Cambridge（剑桥大学） 纠错

Comments arXiv admin note: substantial text overlap with arXiv:2412.00868

Journal ref Forty-second International Conference on Machine Learning. ICML 2025

Diffusion Counterfactual Generation with Semantic Abduction

Rajat Rasal, Avinash Kori, Fabio De Sousa Ribeiro, Tian Xia, Ben Glocker

机构 * Department of Computing, Imperial College London, UK（计算系，伦敦帝国学院，英国） 纠错

Comments Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada

Journal ref PMLR 267, 2025

Identifiable Object Representations under Spatial Ambiguities

Avinash Kori, Francesca Toni, Ben Glocker

机构 * Department of Computing, Imperial College London, London, UK（计算系，伦敦帝国学院，伦敦，英国） 纠错

Journal ref Published as a proceeding of the 42 nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025