ICLR

Federated Domain Generalization with Data-free On-server Matching Gradient

Trong-Binh Nguyen, Minh-Duong Nguyen, Jinsun Park, Quoc-Viet Pham, Won Joo Hwang

机构 * Pusan National University（釜山国立大学） ； Trinity College Dublin（都柏林三一学院） 纠错

Comments 26 pages, 15 figures, ICLR

When is Task Vector Provably Effective for Model Editing? A Generalization Analysis of Nonlinear Transformers

Hongkang Li, Yihua Zhang, Shuai Zhang, Meng Wang, Sijia Liu, Pin-Yu Chen

机构 * Rensselaer Polytechnic Institute（拉特格斯理工学院） ； Michigan State University（密歇根州立大学） ； New Jersey Institute of Technology（新泽西理工学院） ； IBM Research（IBM研究院） 纠错

Comments Published at ICLR 2025 as an oral paper

Societal Impacts Research Requires Benchmarks for Creative Composition Tasks

Judy Hanwen Shen, Carlos Guestrin

Comments v1: ICLR 2025 Workshop on Bidirectional Human-AI Alignment (BiAlign) v2: ICML 2025 Position Paper

Language Models, Graph Searching, and Supervision Adulteration: When More Supervision is Less and How to Make More More

Arvid Frydenlund

机构 * University of Toronto（多伦多大学） ； Vector Institute（向量研究所） 纠错

Comments ACL 2025 Main. A camera-ready version will follow in a few weeks. A reduced version of this work was also accepted to the Workshop on Spurious Correlation and Shortcut Learning: Foundations and Solutions (SCSL) at ICLR 2025

CrossMPT: Cross-attention Message-Passing Transformer for Error Correcting Codes

Seong-Joon Park, Hee-Youl Kwak, Sang-Hyo Kim, Yongjune Kim, Jong-Seon No

机构 * POSTECH（POSTECH大学） ； University of Ulsan（乌山大学） ； Sungkyunkwan University（全北大学） ； Seoul National University（首尔国立大学） 纠错

Comments 17 pages

Journal ref International Conference on Learning Representations, 2025

SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models

Daniel Levy, Siba Smarak Panigrahi, Sékou-Oumar Kaba, Qiang Zhu, Kin Long Kelvin Lee, Mikhail Galkin, Santiago Miret, Siamak Ravanbakhsh

Comments 24 pages, 10 figures, International Conference on Learning Representations (ICLR) 2025

Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Models

Lucas Bandarkar, Benjamin Muller, Pritish Yuvraj, Rui Hou, Nayan Singhal, Hongjiang Lv, Bing Liu

Comments ICLR 2025, Spotlight Paper, In The Thirteenth International Conference on Learning Representations, 2025

Journal ref The Thirteenth International Conference on Learning Representations (2025)

DiffPuter: Empowering Diffusion Models for Missing Data Imputation

Hengrui Zhang, Liancheng Fang, Qitian Wu, Philip S. Yu

Comments ICLR 2025, Spotlight

Fantastic Targets for Concept Erasure in Diffusion Models and Where To Find Them

Anh Bui, Trang Vu, Long Vuong, Trung Le, Paul Montague, Tamas Abraham, Junae Kim, Dinh Phung

Journal ref International Conference on Learning Representations 2025

REvolve: Reward Evolution with Large Language Models using Human Feedback

Rishi Hazra, Alkis Sygkounas, Andreas Persson, Amy Loutfi, Pedro Zuidberg Dos Martires

Comments Published in ICLR 2025. Project page: https://rishihazra.github.io/REvolve

Sufficient conditions for offline reactivation in recurrent neural networks

Nanda H. Krishna, Colin Bredenberg, Daniel Levenstein, Blake A. Richards, Guillaume Lajoie

Comments ICLR 2024

Journal ref The Twelfth International Conference on Learning Representations, 2024

Incorporating Visual Correspondence into Diffusion Model for Virtual Try-On

Siqi Wan, Jingwen Chen, Yingwei Pan, Ting Yao, Tao Mei

Comments ICLR 2025. Code is publicly available at: https://github.com/HiDream-ai/SPM-Diff

Small-to-Large Generalization: Data Influences Models Consistently Across Scale

Alaa Khaddaj, Logan Engstrom, Aleksander Madry

机构 * MIT（麻省理工学院） 纠错

Journal ref ICLR 2025

Last Layer Empirical Bayes

Valentin Villecroze, Yixin Wang, Gabriel Loaiza-Ganem

机构 * University of Michigan（密歇根大学） ； Layer 6 AI 纠错

Comments Accepted at the ICBINB Worshop at ICLR 2025

MLLM as Retriever: Interactively Learning Multimodal Retrieval for Embodied Agents

Junpeng Yue, Xinrun Xu, Börje F. Karlsson, Zongqing Lu

机构 * School of Computer Science, Peking University（北京大学计算机科学学院） ； Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所） ； Beijing Academy of Artificial Intelligence（北京人工智能研究院） 纠错

Comments ICLR 2025

Dataless Knowledge Fusion by Merging Weights of Language Models

Xisen Jin, Xiang Ren, Daniel Preotiuc-Pietro, Pengxiang Cheng

机构 * University of Southern California（南加州大学） ； Bloomberg（布莱尔） 纠错

Comments ICLR 2023; The code is available at https://github.com/bloomberg/dataless-model-merging and https://github.com/AuCson/RegMean-LLama3-8B. Fixed typos

Laplace Sample Information: Data Informativeness Through a Bayesian Lens

Johannes Kaiser, Kristian Schwethelm, Daniel Rueckert, Georgios Kaissis

机构 * Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心） 纠错

Journal ref The Thirteenth International Conference on Learning Representations ICLR (2025)

Learning Novel Skills from Language-Generated Demonstrations

Ao-Qun Jin, Tian-Yu Xiang, Xiao-Hu Zhou, Mei-Jiang Gui, Xiao-Liang Xie, Shi-Qi Liu, Shuang-Yi Wang, Yue Cao, Sheng-Bin Duan, Fu-Chao Xie, Zeng-Guang Hou

机构 * Institute of Automation Chinese Academy of Sciences（自动化研究所，中国科学院） 纠错

Comments 10 pages, International Conference on Learning Representations (ICLR) 2025 Workshop on Generative Models for Robot Learning (GenBot)

On the Performance Analysis of Momentum Method: A Frequency Domain Perspective

Xianliang Li, Jun Luo, Zhiwei Zheng, Hanxiao Wang, Li Luo, Lingkun Wen, Linlong Wu, Sheng Xu

机构 * Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究院） ； University of Chinese Academy of Sciences（中国科学院大学） ； University of California, Berkeley（加州大学伯克利分校） ； Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所） ； Sun Yat-sen University（中山大学） ； Shanghai Astronomical Observatory, Chinese Academy of Sciences（中国科学院上海天文台） ； University of Luxembourg（卢森堡大学） 纠错

Comments ICLR 2025. 22 pages, 14 figures. Keywords: Momentum Method, Stochastic Gradient Descent, Z-Transform, Frequency Domain Analysis, Deep Learning

Local Loss Optimization in the Infinite Width: Stable Parameterization of Predictive Coding Networks and Target Propagation

Satoki Ishikawa, Rio Yokota, Ryo Karakida

机构 * Science Tokyo（东京科学大学） ； AIST（国家信息与通信技术研究所） 纠错

Comments Accepted at ICLR 2025

MediConfusion: Can you trust your AI radiologist? Probing the reliability of multimodal medical foundation models

Mohammad Shahab Sepehri, Zalan Fabian, Maryam Soltanolkotabi, Mahdi Soltanolkotabi

机构 * Department of Electrical and Computer Engineering, University of Southern California（电气与计算机工程系，南加州大学） ； Department of Radiology and Imaging Sciences, University of Utah（放射学与影像科学系，犹他大学） 纠错

Comments 24 Pages, 9 figures, The Thirteenth International Conference on Learning Representations (ICLR) 2025

SANER: Annotation-free Societal Attribute Neutralizer for Debiasing CLIP

Yusuke Hirota, Min-Hung Chen, Chien-Yi Wang, Yuta Nakashima, Yu-Chiang Frank Wang, Ryo Hachiuma

机构 * NVIDIA ； Osaka University（大阪大学） ； National Taiwan University（国立台湾大学） 纠错

Comments ICLR 2025

Learning From Simplicial Data Based on Random Walks and 1D Convolutions

Florian Frantzen, Michael T. Schaub

Journal ref International Conference on Learning Representations 2024

Walk the Talk? Measuring the Faithfulness of Large Language Model Explanations

Katie Matton, Robert Osazuwa Ness, John Guttag, Emre Kıcıman

机构 * MIT（麻省理工学院） ； Microsoft Research（微软研究院） 纠错

Comments 66 pages, 14 figures, 40 tables; ICLR 2025 (spotlight) camera ready

Hotspot-Driven Peptide Design via Multi-Fragment Autoregressive Extension

Jiahan Li, Tong Chen, Shitong Luo, Chaoran Cheng, Jiaqi Guan, Ruihan Guo, Sheng Wang, Ge Liu, Jian Peng, Jianzhu Ma

机构 * Tsinghua University（清华大学） ； University of Washington（华盛顿大学） ； Massachusetts Institute of Technology（麻省理工学院） ； University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） ； ByteDance Inc.（字节跳动公司） ； Helixon Inc.（Helixon公司） 纠错

Comments Published as a conference paper at ICLR 2025

New Evidence of the Two-Phase Learning Dynamics of Neural Networks

Zhanpeng Zhou, Yongyi Yang, Mahito Sugiyama, Junchi Yan

机构 * Shanghai Jiao Tong University（上海交通大学） ； University of Michigan（密歇根大学） ； National Institute of Informatics（国家信息研究所） ； The Graduate University for Advanced Studies, SOKENDAI（SOKENDAI高级研究大学） 纠错

Comments This work extends the workshop paper, On the Cone Effect in the Learning Dynamics, accepted by ICLR 2025 Workshop DeLTa

The Spotlight Resonance Method: Resolving the Alignment of Embedded Activations

George Bird

机构 * Department of Computer Science University of Manchester（计算机科学系 曼彻斯特大学） 纠错

Comments 25 pages, 13 figures, 2nd Workshop on Representational Alignment, International Conference on Learning Representations (ICLR)

Differentially Private Synthetic Data via APIs 3: Using Simulators Instead of Foundation Model

Zinan Lin, Tadas Baltrusaitis, Wenyu Wang, Sergey Yekhanin

机构 * Microsoft Research Redmond（微软研究院红mond分部） ； Microsoft Cambridge（微软剑桥分部） 纠错

Comments Published in: (1) ICLR 2025 Workshop on Data Problems, (2) ICLR 2025 Workshop on Synthetic Data

Perm: A Parametric Representation for Multi-Style 3D Hair Modeling

Chengan He, Xin Sun, Zhixin Shu, Fujun Luan, Sören Pirk, Jorge Alejandro Amador Herrera, Dominik L. Michels, Tuanfeng Y. Wang, Meng Zhang, Holly Rushmeier, Yi Zhou

机构 * Yale University（耶鲁大学） ； Adobe Research（Adobe研究） ； Kiel University（基尔大学） ； KAUST（王国学术科技研究理事会） ； NJUST（南京理工大学） 纠错

Comments Accepted to ICLR 2025. Project page: https://cs.yale.edu/homes/che/projects/perm/

Active Learning on Synthons for Molecular Design

Tom George Grigg, Mason Burlage, Oliver Brook Scott, Adam Taouil, Dominique Sydow, Liam Wilbraham

机构 * Recursion 纠错

Comments 14 pages, 10 figures. Presented at ICLR 2025 GEM Workshop