Novelty Detection in Reinforcement Learning with World Models
Comments ICML Spotlight 2025
期刊&会议
ICML
International Conference on Machine Learning · 会议 · Machine Learning
Improving the Variance of Differentially Private Randomized Experiments through Clustering
Comments 35 pages, 8 figures, accepted at International Conference on Machine Learning (ICML 2025)
Designing Cyclic Peptides via Harmonic SDE with Atom-Bond Modeling
Comments Accepted to ICML 2025
Annealing Flow Generative Models Towards Sampling High-Dimensional and Multi-Modal Distributions
机构 * Dongze Wu ； Yao Xie
Comments This paper has been accepted to ICML 2025 and will appear in the Proceedings of Machine Learning Research (PMLR)
A Cross Modal Knowledge Distillation & Data Augmentation Recipe for Improving Transcriptomics Representations through Morphological Features
Comments ICML 2025 Main Proceedings
Learning Single Index Models with Diffusion Priors
机构 * Anqi Tang（安琪·塔恩） ； Youming Chen（欧阳明·陈） ； Shuchen Xue（苏晨·薛） ； Zhaoqiang Liu（刘兆强）
Comments ICML 2025
TimePro: Efficient Multivariate Long-term Time Series Forecasting with Variable- and Time-Aware Hyper-state
Comments ICML 2025
Open-Det: An Efficient Learning Framework for Open-Ended Detection
Comments ICML 2025
Joint-stochastic-approximation Random Fields with Application to Semi-supervised Learning
Comments ICML 2018 submission. arXiv admin note: text overlap with arXiv:1808.01630, arXiv:2505.18558
Efficient Robotic Policy Learning via Latent Space Backward Planning
Comments Accepted by ICML 2025
The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Analysis of Orthogonal Safety Directions
机构 * Wenbo Pan（温波 Pan） ； Zhichao Liu（刘志超 Liu） ； Qiguang Chen（陈启广 Chen） ； Xiangyang Zhou（周祥阳 Zhou） ； Haining Yu（于航 Yu） ； Xiaohua Jia（贾小华 Jia）
Comments Code and artifacts: https://github.com/BMPixel/safety-residual-space Accepted by ICML 2025
Optimizing Robustness and Accuracy in Mixture of Experts: A Dual-Model Approach
Comments 15 pages, 7 figures, accepted by ICML 2025
Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond
Comments Accepted by ICML 2025
V-CAS: A Realtime Vehicle Anti Collision System Using Vision Transformer on Multi-Camera Streams
机构 * College of EME, NUST（电子工程学院，纳沃伊大学） ； University of Galway（Galway大学）
Comments Accepted at ICMLA 2024
Journal ref In 2024 International Conference on Machine Learning and Applications (ICMLA), pp.939-944
The Best of Both Worlds: Bridging Quality and Diversity in Data Selection with Bipartite Graph
机构 * Monash University（墨尔本大学）
Comments Accepted by ICML 2025
Does Graph Prompt Work? A Data Operation Perspective with Theoretical Analysis
Comments accepted by ICML 2025
Active Inference Meeting Energy-Efficient Control of Parallel and Identical Machines
机构 * Politecnico di Milano（米兰理工学院） ； Rutgers University（罗格斯大学）
Comments Accepted at the 10th International Conference on Machine Learning, Optimization, and Data Science
Diving into Underwater: Segment Anything Model Guided Underwater Salient Instance Segmentation and A Large-scale Dataset
Comments Accepted to ICML 2024, Code released at: https://github.com/LiamLian0727/USIS10K
Flow of Reasoning: Training LLMs for Divergent Reasoning with Minimal Examples
Comments Accepted to ICML 2025
Adaptive Sample Sharing for Multi Agent Linear Bandits
Comments 33 pages
Journal ref Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025
Hierarchical Masked Autoregressive Models with Low-Resolution Token Pivots
Comments ICML 2025. Source code is available at https://github.com/HiDream-ai/himar
Learning Extrapolative Sequence Transformations from Markov Chains
Comments To be published at the Forty-Second International Conference on Machine Learning
Learning Optimal Multimodal Information Bottleneck Representations
Comments ICML 2025
Learning to Trust Bellman Updates: Selective State-Adaptive Regularization for Offline RL
Comments Accepted to ICML 2025
Probabilistic Interactive 3D Segmentation with Hierarchical Neural Processes
Comments ICML 2025 Proceedings
UniMoMo: Unified Generative Modeling of 3D Molecules for De Novo Binder Design
Comments Accepted to ICML 2025
Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)
Comments accepted to ICML 2025 position track
Prediction-Powered E-Values
Comments Accepted at ICML 2025
On the Guidance of Flow Matching
机构 * Department of Artificial Intelligence, Westlake University, Hangzhou, China（人工智能系，西湖大学，杭州，中国） ； Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, China（数学与系统科学研究院，中国科学院，北京，中国）
Comments ICML 2025 spotlight poster. 37 pages, 7 figures
SITCOM: Step-wise Triple-Consistent Diffusion Sampling for Inverse Problems
Journal ref International Conference on Machine Learning 2025
