On the Query Complexity of Verifier-Assisted Language Generation
机构 * Carnegie Mellon University（卡内基梅隆大学） ； Microsoft Research NYC（微软研究院纽约）
Comments ICML 2025
期刊&会议
ICML
International Conference on Machine Learning · 会议 · Machine Learning
Maximum Entropy Reinforcement Learning with Diffusion Policy
Comments ICML 2025
In-context denoising with one-layer transformers: connections between attention and associative memory retrieval
Comments Accepted to ICML 2025
MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding
机构 * Tsinghua University, Beijing, China（清华大学） ； Shanghai Artificial Intelligence Laboratory, Shanghai, China（上海人工智能实验室）
Comments ICML 2025
MapEval: A Map-Based Evaluation of Geo-Spatial Reasoning in Foundation Models
机构 * Department of Computer Science and Engineering Bangladesh University of Engineering and Technology (BUET)（计算机科学与工程系，孟加拉国工程与技术大学） ； Statistics, Islamic University, Bangladesh（统计学，伊斯兰大学，孟加拉国） ； Faculty of Information Technology, Monash University, Melbourne, Australia（信息科技学院，墨尔本大学，澳大利亚） ； Qatar Computing Research Institute (QCRI)（卡塔尔计算研究所）
Comments ICML 2025 (Spotlight)
Training Software Engineering Agents and Verifiers with SWE-Gym
Comments Accepted at ICML 2025. Code at https://github.com/SWE-Gym/SWE-Gym
ResearchTown: Simulator of Human Research Community
机构 * Haofei Yu ； Zhaochen Hong ； Zirui Cheng ； Kunlun Zhu ； Keyang Xuan ； Jinwei Yao ； Tao Feng ； Jiaxuan You
Comments 9 pages, ICML 2025
Diving into Self-Evolving Training for Multimodal Reasoning
Comments ICML 2025, Project Page: https://mstar-lmm.github.io
Reasoning Through Execution: Unifying Process and Outcome Rewards for Code Generation
Comments Accepted to ICML 2025; 23 pages, 7 figures, code is available at: https://github.com/zhuohaoyu/ORPS
The Synergy of LLMs & RL Unlocks Offline Learning of Generalizable Language-Conditioned Policies with Low-fidelity Data
Comments Accepted at International Conference on Machine Learning (ICML) 2025
Normalizing Flows are Capable Generative Models
Comments ICML 2025
Graph Neural Network Generalization with Gaussian Mixture Model Based Augmentation
Journal ref International Conference on Machine Learning (ICML) 2025
SGD Jittering: A Training Strategy for Robust and Accurate Model-Based Architectures
Comments ICML 2025
Ab Initio Nonparametric Variable Selection for Scalable Symbolic Regression with Large $p$
机构 * Shengbin Ye（谢圣彬） ； Meng Li（李萌）
Comments To appear in ICML 2025
PoisonBench: Assessing Large Language Model Vulnerability to Data Poisoning
Comments Accepted at ICML 2025. Tingchen Fu and Fazl Barez are core research contributors
AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML
Comments ICML 2025, Project Page: https://deepauto-ai.github.io/automl-agent
VisionTS: Visual Masked Autoencoders Are Free-Lunch Zero-Shot Time Series Forecasters
Comments v4: accepted by ICML 2025
Banyan: Improved Representation Learning with Explicit Structure
Comments ICML 2025 Camera Ready + Code Release
BoA: Attention-aware Post-training Quantization without Backpropagation
Comments ICML 2025
Certification for Differentially Private Prediction in Gradient-Based Training
Comments ICML 2025. 20 pages, 9 figures
LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Models
Comments In Proceedings of the 42st International Conference on Machine Learning (ICML 2025), Project page at https://ml-research.github.io/human-centered-genai/projects/llavaguard/index.html
Topology-aware Neural Flux Prediction Guided by Physics
Journal ref International Conference on Machine Learning, 2025
Projectable Models: One-Shot Generation of Small Specialized Transformers from Large Ones
机构 * A. Zhmoginov（未知） ； J. Lee（未知） ； M. Sandler（未知）
Comments Presented at ES-FoMo II: 2nd Workshop on Efficient Systems for Foundation Models (ICML 2024)
Bayesian Inference for Correlated Human Experts and Classifiers
Comments accepted to ICML 2025
TabFlex: Scaling Tabular Learning to Millions with Linear Attention
Comments 30 pages, ICML 2025
On Differential Privacy for Adaptively Solving Search Problems via Sketching
Comments ICML 2025
Learning-Augmented Hierarchical Clustering
Comments ICML 2025; abstract shortened for arxiv requirements
Learning-Augmented Algorithms for MTS with Bandit Access to Multiple Predictors
机构 * Bocconi University（博科尼大学）
Comments Accepted to ICML 2025
PCEvolve: Private Contrastive Evolution for Synthetic Dataset Generation via Few-Shot Private Data and Generative APIs
Comments Accepted as ICML Spotlight (top 2.6%)
Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching
Comments Accepted at ICML 2025
