Deep Linear Probe Generators for Weight Space Learning
机构 * School of Computer Science and Engineering（计算机科学与工程学院）
Comments ICLR 2025. Project page: https://vision.huji.ac.il/probegen
期刊&会议
ICLR
International Conference on Learning Representations · 会议 · Machine Learning
LICO: Large Language Models for In-Context Molecular Optimization
机构 * Department of Computer Science University of California, Los Angeles（计算机科学系 加州大学洛杉矶分校）
Comments International Conference on Learning Representations (ICLR 2025)
AtomSurf : Surface Representation for Learning on Protein Structures
机构 * LIX Ecole Polytechnique（LIX 巴黎高等理工学院） ； Mines Paris PSL Research University（巴黎 Mines 工程大学） ； Institut Curie PSL Research University（Curie 研究所 巴黎高等理工大学） ； INSERM U900（国家医学研究院 U900） ； Institute of Bioengineering École Polytechnique Fédérale de Lausanne（洛桑联邦理工学院生物工程研究所）
Comments Published as a conference paper at The Thirteenth International Conference on Learning Representations (ICLR 2025). The official open-access version is available at https://openreview.net/forum?id=ARQIJXFcTH
Journal ref The Thirteenth International Conference on Learning Representations (ICLR), 2025
Topoformer: brain-like topographic organization in Transformer language models through spatial querying and reweighting
机构 * Novus Technologies ； MIT（麻省理工学院） ； Harvard University（哈佛大学）
Comments ICLR 2024 Workshop on Representational Alignment (Re-Align) Camera Ready
Attention-Guided Deep Adversarial Temporal Subspace Clustering (A-DATSC) Model for multivariate spatiotemporal data
机构 * University of Maryland, Baltimore County（马里兰大学巴尔的摩县）
Comments 9 pages, under review submitted to ICLR 2025
Enabling Fine-Grained Operating Points for Black-Box LLMs
机构 * Amazon（亚马逊）
Comments Under review at ICLR 2026. 36 pages, 17 figures
BIRD: A Trustworthy Bayesian Inference Framework for Large Language Models
机构 * University of Pennsylvania（宾夕法尼亚大学） ； Arizona State University（亚利桑那州立大学）
Journal ref ICLR 2025 (Oral)
Generalized Principal-Agent Problem with a Learning Agent
机构 * Harvard University（哈佛大学）
Comments A short version of this work appeared on ICLR 2025 (spotlight). This full version has been accepted by Quantitative Economics
Watermark Robustness and Radioactivity May Be at Odds in Federated Learning
Comments 9 pages, 4 figures (not including citation and appendix) submitted to ICLR 2026
Closing the Curvature Gap: Full Transformer Hessians and Their Implications for Scaling Laws
机构 * Yandex ； BRAIn Lab ； Moscow State University（莫斯科国立大学）
Comments 38 pages, 12 figures. Submitted to ICLR 2026
A Generic Framework for Conformal Fairness
机构 * The Ohio State University（俄亥俄州立大学） ； Dolby Laboratories（杜比实验室） ； Cisco Systems（思科系统）
Comments ICLR 2025 Camera Ready Version
What should a neuron aim for? Designing local objective functions based on information theory
机构 * Faculty of Physics, Institute for the Dynamics of Complex Systems, University of Göttingen（物理学院、复杂系统动力学研究所、哥廷根大学） ； Max Planck Institute for Dynamics and Self-Organization, Göttingen（动态与自我组织马克斯·普朗克研究所，哥廷根） ； Göttingen Campus Institute for Dynamics of Biological Networks, University of Göttingen（哥廷根校区生物网络动力学研究所、哥廷根大学） ； Institute of Computer Science and Campus Institute Data Science, University of Göttingen（计算机科学研究所和校园数据科学研究所、哥廷根大学）
Comments Presented as an oral at ICLR 2025. Conference version: https://openreview.net/forum?id=CLE09ESvul, 24 pages, 11 figures
DRO-InstructZero: Distributionally Robust Prompt Optimization for Large Language Models
机构 * Department of Electrical Engineering and Computer Science（电气工程与计算机科学系） ； Massachusetts Institute of Technology（麻省理工学院）
Comments Preprint. Under review at ICLR 2026. 11 pages, 2 figures
DiffEnc: Variational Diffusion with a Learned Encoder
机构 * Copenhagen University Hospital（哥本哈根大学医院）
Journal ref International Conference on Representation Learning 2024 (ICLR 2024)
Synthesizing Agentic Data for Web Agents with Progressive Difficulty Enhancement Mechanisms
机构 * Salesforce AI Research（Salesforce AI研究院） ； University of Wisconsin-Madison（威斯康星大学麦迪逊分校）
Comments Preprint. ICLR 26 submission
InfoDet: A Dataset for Infographic Element Detection
机构 * BNRist, Tsinghua University（清华大学信息与技术研究院）
Comments Submitted to ICLR 2026
SOHES: Self-supervised Open-world Hierarchical Entity Segmentation
机构 * University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） ； Adobe Research（Adobe研究院）
Comments ICLR 2024
A Simple Framework for Open-Vocabulary Zero-Shot Segmentation
机构 * EPFL（瑞士联邦理工学院） ； KU Leuven（比利时卢旺达大学） ； CHUV（日内瓦大学医院）
Comments ICLR 2025
Do LLM Agents Have Regret? A Case Study in Online Learning and Games
机构 * Massachusetts Institute of Technology（麻省理工学院） ； University of Maryland, College Park（马里兰大学College Park分校）
Comments Camera ready version of ICLR 2025
Inclusive Fitness as a Key Step Towards More Advanced Social Behaviors in Multi-Agent Reinforcement Learning Settings
机构 * Vrije Universiteit Brussel（布鲁塞尔自由大学）
Comments This version is a slightly updated version (e.g., added an important reference) compared to the peer-reviewed versions at 'Adapative Learning Agents' at AAMAS 2022 or 'From Cells to Societies' at ICLR 2022
The Robustness of Differentiable Causal Discovery in Misspecified Scenarios
机构 * School of Mathematics, Southeast University, Nanjing 210096, China（数学学院，东南大学，南京210096，中国） ； School of Cyber Science and Engineering, Southeast University, Nanjing 210096, China（网络科学与工程学院，东南大学，南京210096，中国）
Comments accepted to ICLR 2025
A Set-Sequence Model for Time Series
机构 * Stanford University（斯坦福大学）
Comments Presented at the Workshop on Financial AI at ICLR 2025
Protein as a Second Language for LLMs
机构 * Wuhan University（武汉大学） ； Hong Kong Polytechnic University（香港理工大学） ； Stanford University（斯坦福大学） ； Topify AI
Comments Main paper: 9 pages, 6 figures. With references and appendix: 18 pages, 9 figures total. Submitted to ICLR 2026 (under review)
Judge's Verdict: A Comprehensive Analysis of LLM Judge Capability Through Human Agreement
机构 * NVIDIA Corporation（NVIDIA公司）
Comments 10 pages, 1 figure, 4 tables, under review as a conference paper at ICLR 2026
Are Transformers Able to Reason by Connecting Separated Knowledge in Training Data?
Comments Accepted by ICLR 2025
Can a MISL Fly? Analysis and Ingredients for Mutual Information Skill Learning
机构 * Princeton University（普林斯顿大学）
Comments ICLR 2025 (Oral < 1.8%). Code and videos are available on the website: https://princeton-rl.github.io/contrastive-successor-features/
Human-inspired Episodic Memory for Infinite Context LLMs
机构 * Huawei Noah’s Ark Lab（华为诺亚实验室） ； University College London（伦敦大学学院）
Journal ref Proc. International Conference on Learning Representations (ICLR), 2025
Time-Aware Feature Selection: Adaptive Temporal Masking for Stable Sparse Autoencoder Training
机构 * Yale University（耶鲁大学） ； University of Chicago（芝加哥大学）
Comments First submitted on February 10th, 2025 to ICLR 2025 Workshop (XAI4Science: From Understanding Model Behavior to Discovering New Scientific Knowledge). The paper was accepted but the workshop does not generate proceedings. Now uploading to arXiv to make the paper publicly available
Contrastive Difference Predictive Coding
机构 * Carnegie Mellon University（卡内基梅隆大学） ； Princeton University（普林斯顿大学）
Comments ICLR 2024. Website (https://chongyi-zheng.github.io/td_infonce) and code (https://github.com/chongyi-zheng/td_infonce)
MT-DAO: Multi-Timescale Distributed Adaptive Optimizers with Local Updates
Comments Submitted to the ICLR 2026 Conference
