Prune 'n Predict: Optimizing LLM Decision-making with Conformal Prediction
Journal ref 42nd International Conference on Machine Learning (ICML 2025)
期刊&会议
ICML
International Conference on Machine Learning · 会议 · Machine Learning
On the Local Complexity of Linear Regions in Deep ReLU Networks
机构 * Department of Mathematics, UCLA, USA ； Department of Statistics \& Data Science, UCLA, USA
Comments International Conference on Machine Learning (ICML), 2025
Challenges and Guidelines in Deep Generative Protein Design: Four Case Studies
Comments Proceedings of the ICML 2025 Workshop on Multi-modal Foundation Models and Large Language Models for Life Sciences, Vancouver, Canada. 2025. Copyright 2025 by the author(s)
LASER: Attention with Exponential Transformation
机构 * University of Texas at Austin（德克萨斯大学奥斯汀分校） ； Google（谷歌）
Comments ICML 2025
Leveraging Skills from Unlabeled Prior Data for Efficient Online Exploration
机构 * University of California, Berkeley（加州大学伯克利分校）
Comments 28 pages, 20 figures, 42nd International Conference on Machine Learning (ICML 2025)
Task-Agnostic Pre-training and Task-Guided Fine-tuning for Versatile Diffusion Planner
机构 * Northwestern Polytechnical University（西北工业大学） ； Institute of Artificial Intelligence (TeleAI), China Telecom（人工智能研究院（TeleAI），中国电信） ； Tsinghua University（清华大学） ； Hong Kong University of Science（香港科学大学） ； Shenzhen Research Institute of Northwestern Polytechnical University（西北工业大学深圳研究院）
Comments Published at ICML 2025
Insuring Uninsurable Risks from AI: Government as Insurer of Last Resort
机构 * Independent Researcher, Cambridge Boston Alignment Initiative（独立研究者，剑桥波士顿对齐计划）
Comments Accepted to Generative AI and Law Workshop at the International Conference on Machine Learning (ICML 2024)
Don't be so negative! Score-based Generative Modeling with Oracle-assisted Guidance
机构 * Department of Computer Science, University of British Columbia, Vancouver, Canada ； Montr\'eal Institute for Learning Algorithms (MILA)
Comments ICML 2025 camera ready
Journal ref Proceedings of the 41st International Conference on Machine Learning, PMLR 235:37164-37187, 2024
KELPS: A Framework for Verified Multi-Language Autoformalization via Semantic-Syntactic Alignment
机构 * School of Information Science and Technology, University of Science and Technology of China（信息科学与技术学院，中国科学技术大学） ； USTC Knowledge Computing Lab（中国科学技术大学知识计算实验室）
Comments Accepted by the ICML 2025 AI4MATH Workshop. 22 pages, 16 figures, 2 tables
Open Source Planning & Control System with Language Agents for Autonomous Scientific Discovery
机构 * Google DeepMind（谷歌DeepMind） ； Haverford College（哈弗学院） ； California Institute of Technology（加州理工学院） ； MCML - Munich Center for Machine Learning（慕尼黑机器学习中心） ； University of Sussex（Sussex大学） ； University of Cambridge（剑桥大学） ； Kavli Institute for Cosmology（卡文迪许研究所） ； Institute for Theoretical Particle Physics and Cosmology（理论粒子物理与宇宙学研究所） ； RWTH Aachen University（亚琛工业大学） ； Universitat Autònoma de Barcelona（巴塞罗那自治大学） ； Indian Institute of Science Education and Research（印度科学教育与研究学院） ； Infosys Ltd（Infosys有限公司） ； Flatiron Institute（Flatiron研究所） ； Princeton University（普林斯顿大学） ； Columbia University（哥伦比亚大学） ； University of California, San Diego（加州大学圣地亚哥分校）
Comments Accepted contribution to the ICML 2025 Workshop on Machine Learning for Astrophysics. Code: https://github.com/CMBAgents/cmbagent Videos: https://www.youtube.com/@cmbagent HuggingFace: https://huggingface.co/spaces/astropilot-ai/cmbagent Cloud: https://cmbagent.cloud
Feature learning from non-Gaussian inputs: the case of Independent Component Analysis in high dimensions
机构 * International School of Advanced Studies (SISSA)（国际先进研究学院（SISSA））
Journal ref Oral presentation @ ICML 2025
Feature Learning beyond the Lazy-Rich Dichotomy: Insights from Representational Geometry
机构 * Center for Computational Neuroscience, Flatiron Institute（计算神经科学中心，扁铁研究所） ； University of California, UCLA（加州大学，UCLA） ； Center for Neural Science, New York University（神经科学中心，纽约大学）
Comments This work was published in ICML 2025 and was selected for a spotlight presentation
Open Materials Generation with Stochastic Interpolants
机构 * Center for Soft Matter Research, Department of Physics, New York University, New York 10003, USA ； Simons Center for Computational Physical Chemistry, Department of Chemistry, New York University, New York 10003, USA ； Department of Aerospace Engineering ； Mechanics, University of Minnesota, Minneapolis, MN 55455, USA ； Department of Computer Science \& Engineering, University of Minnesota, Minneapolis, MN 55455, USA ； Department of Physics, University of Florida, Gainesville, FL 32611, USA ； Quantum Theory Project, University of Florida, Gainesville, FL 32611, USA ； Department of Chemistry, University of Florida, Gainesville, FL 32611, USA ； Department of Physics \& Astronomy, Brigham Young University, Provo, UT 84602, USA ； Department of Materials Science \& Engineering, University of Florida, Gainesville, FL 32611, USA ； Courant Institute of Mathematical Sciences, New York University, New York 10003, USA ； Center for Neural Science, New York University, New York 10003, USA
Comments Accepted at Forty-second International Conference on Machine Learning (ICML): https://openreview.net/forum?id=gHGrzxFujU
Online Pre-Training for Offline-to-Online Reinforcement Learning
机构 * LG AI Research, Seoul, Republic of Korea.（LG人工智能研究所在韩国首尔） ； School of Electrical Engineering, KAIST, Daejeon, Republic of Korea（韩国成均馆大学电气工程学院） ； Department of XXX, University of YYY, Location, Country（YYY大学XXX系） ； School of ZZZ, Institute of WWW, Location, Country（WWW研究所ZZZ学院）
Comments ICML 2025 camera-ready
FlowDrag: 3D-aware Drag-based Image Editing with Mesh-guided Deformation Vector Flow Fields
机构 * Korea Advanced Institute of Science and Technology (KAIST)（韩国科学技术院）
Comments ICML 2025 Spotlight
A Practical Two-Stage Recipe for Mathematical LLMs: Maximizing Accuracy with SFT and Efficiency with Reinforcement Learning
机构 * Aillis Inc.（Aillis公司） ； Department of Health Policy（健康政策部门） ； Public Health, Graduate School of Pharmaceutical Sciences, The University of Tokyo, Tokyo, Japan（公共卫生，药学研究生院，东京大学，东京，日本） ； Rist Inc.（Rist公司）
Comments Presented at ICML 2025 Workshop on The second AI for MATH
Raptor: Scalable Train-Free Embeddings for 3D Medical Volumes Leveraging Pretrained 2D Foundation Models
机构 * Computer Science Department, University of California, Los Angeles（计算机科学系，加州大学洛杉矶分校） ； Department of Computational Medicine, David Geffen School of Medicine, University of California, Los Angeles（计算医学系，大卫·格芬医学院，加州大学洛杉矶分校） ； Bioinformatics Interdepartmental Program, University of California, Los Angeles（生物信息学跨学科项目，加州大学洛杉矶分校） ； Department of Human Genetics, University of California, Los Angeles（人类遗传学系，加州大学洛杉矶分校）
Comments 21 pages, 10 figures, accepted to ICML 2025. The first two authors contributed equally
Journal ref In Proc. 42th International Conference on Machine Learning (ICML 2025 Spotlight)
Review, Remask, Refine (R3): Process-Guided Block Diffusion for Text Generation
机构 * Jerome Fisher M\&T Program, University of Pennsylvania, Philadelphia, USA ； Computer Information Sciences, University of Pennsylvania, Philadelphia, USA
Comments Accepted at Methods and Opportunities at Small Scale (MOSS), ICML 2025
UmbraTTS: Adapting Text-to-Speech to Environmental Contexts with Flow Matching
机构 * aiOla Research（aiOla研究）
Comments ICML Workshop on Machine Learning for Audio 2025
Grokking Beyond the Euclidean Norm of Model Parameters
机构 * Mila, Quebec AI Institute, Montréal, Quebec, Canada（魁北克AI研究所） ； CHU Sainte-Justine Research Center, Montréal, Quebec, Canada（圣-朱斯特研究中心） ； CIFAR AI Chair（CIFAR人工智能席位）
Comments 67 pages, 35 figures. Forty-second International Conference on Machine Learning (ICML), 2025
Compositional Risk Minimization
机构 * Meta FAIR
Comments Proceedings of the 42nd International Conference on Machine Learning (ICML) 2025
PAC-Bayes Analysis for Recalibration in Classification
机构 * The University of Osaka, Osaka, Japan（大阪大学） ； RIKEN Center for Advanced Intelligence Project, Tokyo, Japan（理化学研究所先进人工智能项目）
Comments Accepted by the 42nd International Conference on Machine Learning (ICML2025), 38 pages, 8 figures
Pocket2Mol: Efficient Molecular Sampling Based on 3D Protein Pockets
机构 * Tsinghua University, Beijing, China（清华大学） ； HeliXon Limited, Beijing, China（HeliXon有限公司） ； Department of Computer Science, University of Illinois at Urbana-Champaign, Champaign, USA（伊利诺伊大学厄巴纳-香槟分校计算机科学系） ； Westlake University, Hangzhou, China（西湖大学） ； Department of Electronic Engineering, Tsinghua University, Beijing, China（清华大学电子工程系） ； Institute for AI Industry Research, Tsinghua University, Beijing, China（清华大学人工智能产业研究院）
Comments ICML 2022 accepted
Meek Models Shall Inherit the Earth
机构 * MIT CSAIL（麻省理工学院计算机科学与人工智能实验室）
Comments 13 pages, 9 figures, longer version of the paper presented at TAIG ICML 2025
Towards Benchmarking Foundation Models for Tabular Data With Text
机构 * Department of Computer Science, University of Freiburg, Freiburg, Germany（弗赖堡大学计算机科学系） ； Prior Labs, Freiburg, Germany（Prior实验室） ； ELLIS Institute, Tubingen, Germany（图宾根ELLIS研究所）
Comments Accepted at Foundation Models for Structured Data workshop at ICML 2025
Conditional Unigram Tokenization with Parallel Data
机构 * Institute of Formal and Applied Linguistics, Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic（形式与应用语言学研究所，数学与物理系，查尔斯大学，布拉格，捷克共和国）
Comments 21 pages, 4 figures, submitted to Tokenization Workshop (TokShop) at ICML 2025
Distributed and Decentralised Training: Technical Governance Challenges in a Shifting AI Landscape
机构 * Minerva University（明维大学） ； Center for a New American Security（新美国安全中心）
Comments Accepted as an oral presentation at the Technical AI Governance Workshop (ICML 2025)
Sparse Causal Discovery with Generative Intervention for Unsupervised Graph Domain Adaptation
机构 * State Key Laboratory for Multimedia Information Processing, School of Computer Science, PKU-Anker LLM Lab（多媒体信息处理国家重点实验室，计算机科学学院，PKU-Anker LLM实验室） ； Peking University（北京大学） ； University of California, Los Angeles（加州大学洛杉矶分校） ； Paul G. Allen School of Computer Science（保罗·G·艾伦计算机科学学院） ； University of Washington（华盛顿大学） ； Hong Kong University of Science and Technology（香港科技大学）
Comments ICML 2025
Position: We Need An Algorithmic Understanding of Generative AI
机构 * BIFOLD--Berlin Institute for the Foundations of Learning（柏林学习与数据基础研究院） ； University of California Los Angeles（加州大学洛杉矶分校） ； Halıcıoğlu Data Science Institute, University of California San Diego（哈里卡鲁斯数据科学研究所，加州大学圣地亚哥分校） ； Microsoft Research NYC（微软研究院纽约）
Comments Accepted at ICML 2025 as a Spotlight Position Paper
Towards Interpretable Time Series Foundation Models
Comments International Conference on Machine Leaning (ICML) 2025 Workshop on Foundation Models for Structured Data
