arXivDaily arXiv每日学术速递 周一至周五更新

期刊&会议

International Conference on Learning Representations · 会议 · Machine Learning

共收录 9454
2410.07149 2025-04-29 cs.CV cs.LG

Towards Interpreting Visual Information Processing in Vision-Language Models

Clement Neo, Luke Ong, Philip Torr, Mor Geva, David Krueger, Fazl Barez

机构 * Nanyang Technological University(南洋理工大学) University of Oxford(牛津大学) Tel Aviv University(特拉维夫大学) MILA(蒙特利尔人工智能研究院) ERA-Krueger AI Safety Lab(ERA-Krueger人工智能安全实验室)

Comments Published at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2409.06601 2025-04-29 cs.CL cs.LG

LaMsS: When Large Language Models Meet Self-Skepticism

Yetao Wu, Yihong Wang, Teng Chen, Ningyuan Xi, Qingqing Gu, Hongyang Lei, Luo Ji

机构 * Geely AI Lab(Geely人工智能实验室) Beihang University(北京航空航天大学)

Comments 11 pages, 6 figures, ICLR 2025 Workshop SSI-FM,

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.18160 2025-04-28 cs.LG cs.AI cs.RO

Offline Learning of Controllable Diverse Behaviors

Mathieu Petitbois, Rémy Portelas, Sylvain Lamprier, Ludovic Denoyer

机构 * Ubisoft La Forge(育碧La Forge) University of Angers(昂西大学) H Company(H公司)

Comments Generative Models for Robot Learning Workshop at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.18147 2025-04-28 cs.CR cs.LG

NoEsis: Differentially Private Knowledge Transfer in Modular LLM Adaptation

Rob Romijnders, Stefanos Laskaridis, Ali Shahin Shamsabadi, Hamed Haddadi

机构 * Brave Research(Brave研究)

Comments ICLR 2025 MCDC workshop

Journal ref ICLR 2025 Workshop on Modularity for Collaborative, Decentralized, and Continual Deep Learning

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.18113 2025-04-28 cs.LG cs.AI

Learning from Less: SINDy Surrogates in RL

Aniket Dixit, Muhammad Ibrahim Khan, Faizan Ahmed, James Brusey

机构 * Coventry University(科文特大学)

Comments World Models @ ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.10894 2025-04-28 cs.CL cs.AI cs.LG

HyperDAS: Towards Automating Mechanistic Interpretability with Hypernetworks

Jiuding Sun, Jing Huang, Sidharth Baskaran, Karel D'Oosterlinck, Christopher Potts, Michael Sklar, Atticus Geiger

机构 * Stanford University(斯坦福大学) Confirm Labs(确认实验室) Ghent University(根特大学)

Comments ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.21669 2025-04-28 cs.CV

Investigating Memorization in Video Diffusion Models

Chen Chen, Enhuai Liu, Daochang Liu, Mubarak Shah, Chang Xu

机构 * School of Computer Science, Faculty of Engineering, The University of Sydney, Australia(悉尼大学计算机科学学院,工程学院) School of Physics, Mathematics and Computing, The University of Western Australia, Australia(西澳大学物理、数学与计算学院) Center for Research in Computer Vision, University of Central Florida, USA(佛罗里达中央大学计算机视觉研究中心)

Comments Accepted at DATA-FM Workshop @ ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.21665 2025-04-28 cs.CV

Exploring Local Memorization in Diffusion Models via Bright Ending Attention

Chen Chen, Daochang Liu, Mubarak Shah, Chang Xu

机构 * School of Computer Science, Faculty of Engineering, The University of Sydney, Australia(悉尼大学计算机科学学院、工程学院) School of Physics, Mathematics and Computing, The University of Western Australia, Australia(西澳大学物理、数学与计算学院) Center for Research in Computer Vision, University of Central Florida, USA(佛罗里达中央大学计算机视觉研究中心)

Comments Accepted at ICLR 2025 (Spotlight). Project page: https://chenchen-usyd.github.io/BE-Project-Page/

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.03727 2025-04-28 cs.CL cs.AI cs.LG

FaithEval: Can Your Language Model Stay Faithful to Context, Even If "The Moon is Made of Marshmallows"

Yifei Ming, Senthil Purushwalkam, Shrey Pandit, Zixuan Ke, Xuan-Phi Nguyen, Caiming Xiong, Shafiq Joty

机构 * Salesforce AI Research(Salesforce AI研究院) University of Texas at Austin(德克萨斯大学奥斯汀分校)

Comments The conference version of this paper is published at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2409.08813 2025-04-28 cs.CL

Your Weak LLM is Secretly a Strong Teacher for Alignment

Leitian Tao, Yixuan Li

机构 * Department of Computer Sciences, University of Wisconsin-Madison(计算机科学系,威斯康星大学麦迪逊分校)

Comments Accepted by ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2408.06621 2025-04-28 cs.LG cs.CL

Towards Robust and Parameter-Efficient Knowledge Unlearning for LLMs

Sungmin Cha, Sungjun Cho, Dasol Hwang, Moontae Lee

机构 * New York University(纽约大学) University of Wisconsin-Madison(威斯康星大学麦迪逊分校) LG AI Research(LG人工智能研究) University of Illinois Chicago(伊利诺伊大学芝加哥分校)

Comments ICLR 2025 camera-ready version

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.17321 2025-04-25 physics.geo-ph cs.LG

Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space

Michael J. Smith, Luke Fleming, James E. Geach, Ryan J. Roberts, Freddie Kalaitzis, James Banister

机构 * Aspia Space(Aspia空间)

Comments 9 pages, 6 figures, spotlight at `Tackling Climate Change with Machine Learning', ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.17309 2025-04-25 cs.CL

CoheMark: A Novel Sentence-Level Watermark for Enhanced Text Quality

Junyan Zhang, Shuliang Liu, Aiwei Liu, Yubo Gao, Jungang Li, Xiaojie Gu, Xuming Hu

机构 * The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州)) Tsinghua University(清华大学)

Comments Published at the 1st workshop on GenAI Watermarking, collocated with ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.02703 2025-04-25 cs.CL cs.AI cs.LG

Selective Attention Improves Transformer

Yaniv Leviathan, Matan Kalman, Yossi Matias

机构 * Google Research(谷歌研究)

Comments ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2409.19151 2025-04-25 cs.CL

Can LLMs Really Learn to Translate a Low-Resource Language from One Grammar Book?

Seth Aycock, David Stap, Di Wu, Christof Monz, Khalil Sima'an

机构 * Institute for Logic, Language and Computation(逻辑、语言与计算研究所) Language Technology Lab(语言技术实验室) University of Amsterdam(阿姆斯特丹大学)

Comments Accepted at ICLR 2025 (Spotlight)

详情

展开后加载摘要…

URL PDF HTML 收藏
2409.11242 2025-04-25 cs.CL

Measuring and Enhancing Trustworthiness of LLMs in RAG through Grounded Attributions and Learning to Refuse

Maojia Song, Shang Hong Sim, Rishabh Bhardwaj, Hai Leong Chieu, Navonil Majumder, Soujanya Poria

机构 * Singapore University of Technology and Design(新加坡科技设计大学) DSO National Laboratories(DSO国家实验室)

Comments Published at ICLR 2025 (Oral)

详情

展开后加载摘要…

URL PDF HTML 收藏
2408.14837 2025-04-25 cs.LG cs.AI cs.CV

Diffusion Models Are Real-Time Game Engines

Dani Valevski, Yaniv Leviathan, Moab Arar, Shlomi Fruchter

机构 * Google Research(谷歌研究) Google DeepMind(谷歌DeepMind) Tel Aviv University(特拉维夫大学)

Comments ICLR 2025. Project page: https://gamengen.github.io/

详情

展开后加载摘要…

URL PDF HTML 收藏
2405.12519 2025-04-25 cs.LG cs.AI q-bio.QM

MAGE: Model-Level Graph Neural Networks Explanations via Motif-based Graph Generation

Zhaoning Yu, Hongyang Gao

机构 * Department of Computer Science(计算机科学系) Iowa State University(爱荷华州立大学)

Comments arXiv admin note: text overlap with arXiv:2405.08419 The Thirteenth International Conference on Learning Representations 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2406.04724 2025-04-25 cs.LG cs.AI

On Minimizing Adversarial Counterfactual Error in Adversarial RL

Roman Belaire, Arunesh Sinha, Pradeep Varakantham

机构 * Singapore Management University(新加坡国立管理学院) Rutgers University(罗格斯大学)

Comments Presented at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.16929 2025-04-24 cs.LG cs.AI cs.CV cs.IT math.IT

I-Con: A Unifying Framework for Representation Learning

Shaden Alshammari, John Hershey, Axel Feldmann, William T. Freeman, Mark Hamilton

机构 * MIT(麻省理工学院) Google(谷歌) Microsoft(微软)

Comments ICLR 2025; website: https://aka.ms/i-con . Proceedings of the Thirteenth International Conference on Learning Representations (ICLR 2025)

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.16886 2025-04-24 q-bio.QM cs.LG q-bio.BM

Exploring zero-shot structure-based protein fitness prediction

Arnav Sharma, Anthony Gitter

机构 * Department of Computer Sciences(计算机科学系) Department of Biostatistics and Medical Informatics(生物统计学与医学信息学系) Morgridge Institute for Research(Morgridge研究所)

Comments 26 pages, 7 figures

Journal ref ICLR 2025 Workshop on Generative and Experimental Perspectives for Biomolecular Design

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.16864 2025-04-24 stat.ME cs.LG econ.EM stat.ML

Common Functional Decompositions Can Mis-attribute Differences in Outcomes Between Populations

Manuel Quintero, William T. Stephenson, Advik Shreekumar, Tamara Broderick

机构 * MIT IDSS(MIT工业系统科学系) MIT Lincoln Laboratory(MIT林肯实验室) MIT Economics(MIT经济学系) MIT EECS(MIT电子工程与计算机科学系)

Comments 30 pages, appearing in 2nd Workshop on Navigating and Addressing Data Problems for Foundation Models (DATA-FM @ ICLR 2025)

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.19215 2025-04-24 cs.CV

On Symmetries in Convolutional Weights

Bilal Alsallakh, Timothy Wroge, Vivek Miglani, Narine Kokhlikyan

机构 * Voxel AI Labs(Voxel AI实验室) Meta AI

Comments Accepted to the ICLR 2025 Workshop on Weight Space Learning (WSL)

详情

展开后加载摘要…

URL PDF HTML 收藏
2411.05000 2025-04-24 cs.CL

Needle Threading: Can LLMs Follow Threads through Near-Million-Scale Haystacks?

Jonathan Roberts, Kai Han, Samuel Albanie

机构 * University of Cambridge(剑桥大学) The University of Hong Kong(香港大学)

Comments Accepted at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.06172 2025-04-24 cs.AI cs.CL

Multimodal Situational Safety

Kaiwen Zhou, Chengzhi Liu, Xuandong Zhao, Anderson Compalas, Dawn Song, Xin Eric Wang

机构 * University of California, Santa Cruz(加州大学圣克ruz分校) University of California, Berkeley(加州大学伯克利分校)

Comments ICLR 2025 Camera Ready

详情

展开后加载摘要…

URL PDF HTML 收藏
2407.16615 2025-04-24 cs.CL cs.AI cs.LG

Lawma: The Power of Specialization for Legal Annotation

Ricardo Dominguez-Olmedo, Vedant Nanda, Rediet Abebe, Stefan Bechtold, Christoph Engel, Jens Frankenreiter, Krishna Gummadi, Moritz Hardt, Michael Livermore

机构 * Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所) Tübingen AI Center(图宾根人工智能中心) Max Planck Institute for Software Systems(马克斯·普朗克软件系统研究所) ETH Zurich(苏黎世联邦理工学院) Max Planck Institute for Research on Collective Goods(马克斯·普朗克集体研究研究所) Washington University in St. Louis - School of Law(圣路易斯华盛顿大学法学院) University of Virginia School of Law(弗吉尼亚大学法学院)

Comments ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2406.00622 2025-04-24 cs.CV cs.AI

Compositional 4D Dynamic Scenes Understanding with Physics Priors for Video Question Answering

Xingrui Wang, Wufei Ma, Angtian Wang, Shuo Chen, Adam Kortylewski, Alan Yuille

机构 * Johns Hopkins University(约翰霍普金斯大学) Tsinghua University(清华大学) Max Planck Institute for Informatics(马克斯·普朗克研究所(信息学)) University of Freiburg(弗赖堡大学)

Comments ICLR 2025 accepted paper. Project url: https://xingruiwang.github.io/projects/DynSuperCLEVR/

详情

展开后加载摘要…

URL PDF HTML 收藏
2403.12766 2025-04-24 cs.CL

NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens

Cunxiang Wang, Ruoxi Ning, Boqi Pan, Tonghui Wu, Qipeng Guo, Cheng Deng, Guangsheng Bao, Xiangkun Hu, Zheng Zhang, Qian Wang, Yue Zhang

机构 * Westlake University(西湖大学) University of Waterloo(多伦多大学) Hangzhou Normal University(杭州师范大学) Shanghai AI Lab(上海人工智能实验室) SJTU(上海交通大学) NYU Shanghai(纽约大学上海分校)

Comments Accepted by ICLR-2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2504.16053 2025-04-23 cs.CL cs.AI

LongMamba: Enhancing Mamba's Long Context Capabilities via Training-Free Receptive Field Enlargement

Zhifan Ye, Kejing Xia, Yonggan Fu, Xin Dong, Jihoon Hong, Xiangchi Yuan, Shizhe Diao, Jan Kautz, Pavlo Molchanov, Yingyan Celine Lin

机构 * Georgia Institute of Technology(佐治亚理工学院) NVIDIA(英伟达)

Comments Accepted by ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2502.11986 2025-04-23 cs.LG

Selective Task Group Updates for Multi-Task Optimization

Wooseong Jeong, Kuk-Jin Yoon

机构 * Korea Advanced Institute of Science and Technology(韩国科学技术院)

Comments Accepted at ICLR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏