ICML

Causal Evidence for the Primordiality of Colors in Trans-Neptunian Objects

Benjamin L. Davis, Mohamad Ali-Dib, Yujia Zheng, Zehao Jin, Kun Zhang, Andrea Valerio Macciò

机构 * Center for Astrophysics and Space Science (CASS)（天文与空间科学中心） ； Carnegie Mellon University（卡内基梅隆大学） ； Center for Astronomy and Astrophysics and Department of Physics（天文与天体物理中心） ； Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学） 纠错

Comments Accepted to ML4Astro 2025 (Machine Learning for Astrophysics workshop at ICML 2025)

Scaling Collapse Reveals Universal Dynamics in Compute-Optimally Trained Neural Networks

Shikai Qiu, Lechao Xiao, Andrew Gordon Wilson, Jeffrey Pennington, Atish Agarwala

机构 * New York University（纽约大学） ； Google DeepMind（谷歌DeepMind） 纠错

Comments ICML 25. Code available at https://github.com/shikaiqiu/supercollapse

Improving Rationality in the Reasoning Process of Language Models through Self-playing Game

Pinzheng Wang, Juntao Li, Zecheng Tang, Haijia Gui, Min zhang

机构 * School of Computer Science and Technology, Soochow University（苏州大学计算机科学与技术学院） ； Key Laboratory of Data Intelligence and Advanced Computing, Soochow University（苏州大学数据智能与先进计算重点实验室） 纠错

Comments Accepted by ICML 2025

Graph Inverse Style Transfer for Counterfactual Explainability

Bardh Prenkaj, Efstratios Zaradoukas, Gjergji Kasneci

机构 * Technical University of Munich, Germany（慕尼黑技术大学） ； Sapienza University of Rome, Italy（罗马萨皮恩扎大学） 纠错

Comments Accepted to ICML'25

PENCIL: Long Thoughts with Short Memory

Chenxiao Yang, Nathan Srebro, David McAllester, Zhiyuan Li

机构 * Toyota Technological Institute at Chicago（丰田技术研究所（芝加哥）） 纠错

Comments Accepted to ICML 2025. Codes in https://github.com/chr26195/PENCIL

Fine-Grained Captioning of Long Videos through Scene Graph Consolidation

Sanghyeok Chu, Seonguk Seo, Bohyung Han

机构 * ECE, Seoul National University, Korea.（电子工程系，首尔国立大学，韩国） ； IPAI, Seoul National University, Korea.（人工智能研究所，首尔国立大学，韩国） 纠错

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

Towards Cost-Effective Reward Guided Text Generation

Ahmad Rashid, Ruotian Wu, Rongqi Fan, Hongliang Li, Agustinus Kristiadi, Pascal Poupart

机构 * University of Waterloo（滑铁卢大学） ； Vector Institute（向量研究所） ； Huawei Technologies（华为技术） 纠错

Comments 18 pages. Work accepted at ICML 2025

Enhancing Long Video Generation Consistency without Tuning

Xingyao Li, Fengzhuo Zhang, Jiachun Pan, Yunlong Hou, Vincent Y. F. Tan, Zhuoran Yang

机构 * National University of Singapore（新加坡国立大学） ； Yale University（耶鲁大学） 纠错

Comments ICML 2025 Workshop on Building Physically Plausible World Models (Best Paper), 32 pages, 17 figures

Toward Robust Hyper-Detailed Image Captioning: A Multiagent Approach and Dual Evaluation Metrics for Factuality and Coverage

Saehyung Lee, Seunghyun Yoon, Trung Bui, Jing Shi, Sungroh Yoon

机构 * Department of Electrical and Computer Engineering, Seoul National University（电气与计算机工程系，首尔国立大学） ； Interdisciplinary Program in Artificial Intelligence, Seoul National University（人工智能交叉学科项目，首尔国立大学） ； Adobe Research（Adobe研究） 纠错

Comments ICML 2025

A Causal World Model Underlying Next Token Prediction: Exploring GPT in a Controlled Environment

Raanan Y. Rohekar, Yaniv Gurwicz, Sungduk Yu, Estelle Aflalo, Vasudev Lal

机构 * Intel Labs（英特尔实验室） 纠错

Comments International Conference on Machine Learning (ICML), 2025

Self-Consistency Preference Optimization

Archiki Prasad, Weizhe Yuan, Richard Yuanzhe Pang, Jing Xu, Maryam Fazel-Zarandi, Mohit Bansal, Sainbayar Sukhbaatar, Jason Weston, Jane Yu

机构 * Meta FAIR ； New York University（纽约大学） 纠错

Comments ICML 2025 (camera-ready)

Collapse-Proof Non-Contrastive Self-Supervised Learning

Emanuele Sansone, Tim Lebailly, Tinne Tuytelaars

Comments ICML 2025

Symmetry-Robust 3D Orientation Estimation

Christopher Scarvelis, David Benhaim, Paul Zhang

机构 * MIT CSAIL 纠错

Comments ICML 2025

Scaling Value Iteration Networks to 5000 Layers for Extreme Long-Term Planning

Yuhui Wang, Qingyuan Wu, Dylan R. Ashley, Francesco Faccio, Weida Li, Chao Huang, Jürgen Schmidhuber

机构 * Center of Excellence for Generative AI, King Abdullah University of Science and Technology（生成人工智能卓越中心，国王 Abdullah 科学与技术大学） ； The Swiss AI Lab IDSIA/USI/SUPSI, Switzerland（瑞士人工智能实验室 IDSIA/USI/SUPSI，瑞士） ； The University of Southampton（南安普顿大学） ； National University of Singapore（新加坡国立大学） 纠错

Comments ICML 2025

Learning from Topology: Cosmological Parameter Estimation from the Large-scale Structure

Jacky H. T. Yip, Adam Rouhiainen, Gary Shiu

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

Comments 7 pages, 4 figures. Accepted to the Synergy of Scientific and Machine Learning Modeling Workshop (ICML 2023) and for publication in Machine Learning: Science and Technology

Journal ref Mach. Learn.: Sci. Technol. 6 025063 (2025)

Kinetic Langevin Diffusion for Crystalline Materials Generation

François Cornet, Federico Bergamin, Arghya Bhowmik, Juan Maria Garcia Lastra, Jes Frellsen, Mikkel N. Schmidt

机构 * Technical University of Denmark（丹麦技术大学） ； Pioneer Center for Artificial Intelligence（先锋人工智能中心） 纠错

Comments Accepted at ICML 2025

Set Valued Predictions For Robust Domain Generalization

Ron Tsibulsky, Daniel Nevo, Uri Shalit

机构 * Department of Computer Science, Tel Aviv University, Tel Aviv, Israel（特拉维夫大学计算机科学系） ； Department of Statistics and Operations Research, Tel Aviv University, Tel Aviv, Israel（特拉维夫大学统计与运筹学系） ； Department of Data and Decisions Science, Technion, Haifa, Israel（技术学院数据与决策科学系） 纠错

Comments ICML 2025

Lion Cub: Minimizing Communication Overhead in Distributed Lion

Satoki Ishikawa, Tal Ben-Nun, Brian Van Essen, Rio Yokota, Nikoli Dryden

机构 * Institute of Science Tokyo（东京科学研究所） ； Lawrence Livermore National Laboratory（劳伦斯利弗莫尔国家实验室） 纠错

Comments ICML 2025 Workshop "Tiny Titans: The Next Wave of On-Device Learning for Foundational Models (TTODLer-FM)"

Fluid Democracy in Federated Data Aggregation

Aditya Vema Reddy Kesari, Krishna Reddy Kesari

机构 * Amazon, US（亚马逊公司） 纠错

Comments ICML 2025 Workshop on Collaborative and Federated Agentic Workflows

Guided Generation for Developable Antibodies

Siqi Zhao, Joshua Moller, Porfi Quintero-Cadena, Lood van Niekerk

Comments Published in ICML 2025 GenBio Workshop

MPF: Aligning and Debiasing Language Models post Deployment via Multi Perspective Fusion

Xin Guan, PeiHsin Lin, Zekun Wu, Ze Wang, Ruibo Zhang, Emre Kazim, Adriano Koshiyama

机构 * University College London（伦敦大学学院） ； Center for long-term AI（长期人工智能中心） 纠错

Comments Accepted at ICML 2025 AIW Workshop

Machine Learning for Proactive Groundwater Management: Early Warning and Resource Allocation

Chuan Li, Ruoxuan Yang

机构 * LIPADE, Université Paris Cité（巴黎大学LIPADE） ； Sorbonne Université（索邦大学） ； SAMOVAR, Telecom SudParis, Institut Polytechnique de Paris（Telecom SudParis 高等理工大学） ； Telecom Paris, Institut Polytechnique de Paris（Telecom Paris 高等理工大学） ； Shanghai Jiao Tong University（上海交通大学） 纠错

Journal ref ICML 2025 Workshop on New in ML, Vancouver Canada, July 13-19 2025

Gateformer: Advancing Multivariate Time Series Forecasting through Temporal and Variate-Wise Attention with Gated Representations

Yu-Hsiang Lan, Eric K. Oermann

机构 * New York University, New York, NY, USA（纽约大学） ； Department of Neurosurgery, NYU Langone Health, New York, NY, USA（神经外科部门，NYU Langone健康） 纠错

Comments Accepted at ICML Workshop on Foundation Models for Structured Data

Is Complex Query Answering Really Complex?

Cosimo Gregucci, Bo Xiong, Daniel Hernandez, Lorenzo Loconte, Pasquale Minervini, Steffen Staab, Antonio Vergari

机构 * Institute for Artificial Intelligence, University of Stuttgart, Germany（人工智能研究所，斯图加特大学） ； School of Informatics University of Edinburgh Edinburgh, UK（信息学院，爱丁堡大学） ； University of Southampton, UK（南安普顿大学） ； Stanford University（斯坦福大学） 纠错

Comments ICML 2025

Reliable algorithm selection for machine learning-guided design

Clara Fannjiang, Ji Won Park

机构 * Prescient Design, Genentech（普森设计，基因泰克） 纠错

Comments ICML 2025

Neural Graph Matching Improves Retrieval Augmented Generation in Molecular Machine Learning

Runzhong Wang, Rui-Xi Wang, Mrunali Manjrekar, Connor W. Coley

机构 * Massachusetts Institute of Technology, Cambridge, MA, United States（麻省理工学院） 纠错

Comments ICML 2025

FeatSharp: Your Vision Model Features, Sharper

Mike Ranzinger, Greg Heinrich, Pavlo Molchanov, Jan Kautz, Bryan Catanzaro, Andrew Tao

机构 * NVIDIA 纠错

Comments ICML 2025 Version

Quantifying the Importance of Data Alignment in Downstream Model Performance

Krrish Chawla, Aryan Sahai, Mario DePavia, Sudharsan Sundar, Brando Miranda, Elyas Obbad, Sanmi Koyejo

机构 * stanford（斯坦福大学） 纠错

Journal ref ICLR DMLR Data-centric Machine Learning Research (2024), ICML DataWorld (2025)

De-mark: Watermark Removal in Large Language Models

Ruibo Chen, Yihan Wu, Junfeng Guo, Heng Huang

机构 * University of Maryland, College Park（马里兰大学 College Park 分校） 纠错

Comments ICML 2025

Breaking the $n^{1.5}$ Additive Error Barrier for Private and Efficient Graph Sparsification via Private Expander Decomposition

Anders Aamand, Justin Y. Chen, Mina Dalirrooyfard, Slobodan Mitrović, Yuriy Nevmyvaka, Sandeep Silwal, Yinzhan Xu

Comments ICML 2025