NeurIPS - arXivDaily 期刊&会议

2509.18942 2025-10-23 cs.AI

Data Efficient Adaptation in Large Language Models via Continuous Low-Rank Fine-Tuning

Xiao Han, Zimo Zhao, Wanyu Wang, Maolin Wang, Zitao Liu, Yi Chang, Xiangyu Zhao

机构 * Zhejiang University of Technology（浙江工业大学）； City University of Hong Kong（香港城市大学）； Jinan University（济南大学）； Jilin University（吉林大学）

Comments Accepted by NeurIPS 2025

2509.01254 2025-10-23 cs.LG

What Expressivity Theory Misses: Message Passing Complexity for GNNs

Niklas Kemper, Tom Wollschläger, Stephan Günnemann

机构 * School of Computation, Information and Technology & Munich Data Science Institute（计算、信息与技术学院及慕尼黑数据科学研究所）

Comments NeurIPS 2025, Spotlight

2508.00890 2025-10-23 cs.AI cs.CL cs.LG

AgentTTS: Large Language Model Agent for Test-time Compute-optimal Scaling Strategy in Complex Tasks

Fali Wang, Hui Liu, Zhenwei Dai, Jingying Zeng, Zhiwei Zhang, Zongyu Wu, Chen Luo, Zhen Li, Xianfeng Tang, Qi He, Suhang Wang

机构 * The Pennsylvania State University（宾夕法尼亚州立大学）； Amazon（亚马逊）

Comments Accepted by NeurIPS 2025

2506.23434 2025-10-23 cs.CV cs.RO

Towards foundational LiDAR world models with efficient latent flow matching

Tianran Liu, Shengwen Zhao, Nicholas Rhinehart

机构 * University of Toronto（多伦多大学）

Comments Accepted to the Thirty-Ninth Conference on Neural Information Processing Systems (NeurIPS 2025), 25 pages, 13 figures

2506.16895 2025-10-23 cs.CV cs.AI cs.LG

With Limited Data for Multimodal Alignment, Let the STRUCTURE Guide You

Fabian Gröger, Shuo Wen, Huyen Le, Maria Brbić

机构 * EPFL（瑞士联邦理工学院）； University of Basel（巴塞尔大学）； HSLU（苏黎世联邦理工学院）

Comments NeurIPS 2025 camera-ready

2506.12323 2025-10-23 cs.CV

Doctor Approved: Generating Medically Accurate Skin Disease Images through AI-Expert Feedback

Janet Wang, Yunbei Zhang, Zhengming Ding, Jihun Hamm

机构 * Tulane University（路易斯安那州立大学）

Comments NeurIPS 2025

2506.11153 2025-10-23 cs.SE cs.LG

QiMeng-MuPa: Mutual-Supervised Learning for Sequential-to-Parallel Code Translation

Changxin Ke, Rui Zhang, Shuo Wang, Li Ding, Guangli Li, Yuanbo Wen, Shuoming Zhang, Ruiyuan Xu, Jin Qin, Jiaming Guo, Chenxi Wang, Ling Li, Qi Guo, Yunji Chen

Comments Accepted to NeurIPS'2025

2506.08616 2025-10-23 math.ST cs.LG stat.ML stat.TH

Generalizing while preserving monotonicity in comparison-based preference learning models

Julien Fageot, Peva Blanchard, Gilles Bareilles, Lê-Nguyên Hoang

机构 * Tournesol（图努索尔）； Kleis Technology（克莱斯技术）； CTU in Prague（布拉格CTU）； Calicarpa（卡利卡尔帕）

Comments Accepted at Neurips 2025

2506.05735 2025-10-23 cs.CL cs.LG

Do LLMs Really Forget? Evaluating Unlearning with Knowledge Correlation and Confidence Awareness

Rongzhe Wei, Peizhi Niu, Hans Hao-Hsun Hsu, Ruihan Wu, Haoteng Yin, Mohsen Ghassemi, Yifan Li, Vamsi K. Potluru, Eli Chien, Kamalika Chaudhuri, Olgica Milenkovic, Pan Li

Comments NeurIPS Camera-Ready Version. Code available at: https://github.com/Graph-COM/Knowledge_Unlearning

2506.04168 2025-10-23 cs.LG cs.AI

Horizon Reduction Makes RL Scalable

Seohong Park, Kevin Frans, Deepinder Mann, Benjamin Eysenbach, Aviral Kumar, Sergey Levine

机构 * University of California, Berkeley（加州大学伯克利分校）； Princeton University（普林斯顿大学）； Carnegie Mellon University（卡内基梅隆大学）

Comments NeurIPS 2025

2506.01162 2025-10-23 cs.DS cs.CR cs.LG stat.ML

Nearly-Linear Time Private Hypothesis Selection with the Optimal Approximation Factor

Maryam Aliakbarpour, Zhan Shi, Ria Stevens, Vincent X. Wang

机构 * Department of Computer Science & Ken Kennedy Institute（计算机科学系及肯尼迪研究所）； Rice University（里士满大学）

Comments 33 pages; to appear in NeurIPS 2025

2506.00711 2025-10-23 cs.LG cs.AI cs.CV

QoQ-Med: Building Multimodal Clinical Foundation Models with Domain-Aware GRPO Training

Wei Dai, Peilin Chen, Chanakya Ekbote, Paul Pu Liang

机构 * MIT Media Lab（MIT媒体实验室）； MIT EECS（MIT电子工程与计算机科学系）

Comments Accepted as Oral at NeurIPS 2025. Revision after camera ready

2505.24625 2025-10-23 cs.CV cs.AI

Learning from Videos for 3D World: Enhancing MLLMs with 3D Vision Geometry Priors

Duo Zheng, Shijia Huang, Yanyang Li, Liwei Wang

机构 * The Chinese University of Hong Kong（香港中文大学）

Comments Accepted by NeurIPS 2025

2505.24379 2025-10-23 cs.LG cs.AI cs.CL cs.CR

Unlearned but Not Forgotten: Data Extraction after Exact Unlearning in LLM

Xiaoyu Wu, Yifei Pang, Terrance Liu, Zhiwei Steven Wu

机构 * Rice University（里士大学）； Carnegie Mellon University（卡内基梅隆大学）

Comments Accepted by Neurips 2025

2505.19955 2025-10-23 cs.LG cs.AI cs.CL

MLR-Bench: Evaluating AI Agents on Open-Ended Machine Learning Research

Hui Chen, Miao Xiong, Yujie Lu, Wei Han, Ailin Deng, Yufei He, Jiaying Wu, Yibo Li, Yue Liu, Bryan Hooi

机构 * National University of Singapore（新加坡国立大学）； University of California, Santa Barbara（加州大学圣巴巴拉分校）； Singapore University of Technology and Design（新加坡科技设计大学）

Comments 49 pages, 9 figures. Accepted by NeurIPS 2025 D&B Track

2505.18871 2025-10-23 stat.ML cs.LG

Non-Stationary Lipschitz Bandits

Nicolas Nguyen, Solenne Gaucher, Claire Vernade

机构 * University of Tübingen（图宾根大学）； École Polytechnique, CMAP（巴黎政治学院（CMAP））

Journal ref NeurIPS 2025

2505.18817 2025-10-23 physics.comp-ph cs.AI

High-order Equivariant Flow Matching for Density Functional Theory Hamiltonian Prediction

Seongsu Kim, Nayoung Kim, Dongwoo Kim, Sungsoo Ahn

机构 * KAIST（韩国科学技术院）； POSTECH

Comments NeurIPS 2025 Spotlight

2505.18524 2025-10-23 cs.CL

metaTextGrad: Automatically optimizing language model optimizers

Guowei Xu, Mert Yuksekgonul, Carlos Guestrin, James Zou

机构 * Tsinghua University（清华大学）； Stanford University（斯坦福大学）

Comments 21 pages, 2 figures

Journal ref NeurIPS 2025

2505.18495 2025-10-23 cs.LG

Beyond Masked and Unmasked: Discrete Diffusion Models via Partial Masking

Chen-Hao Chao, Wei-Fang Sun, Hanwen Liang, Chun-Yi Lee, Rahul G. Krishnan

机构 * University of Toronto, Vector Institute（多伦多大学、向量研究所）； NVIDIA AI Technology Center（NVIDIA人工智能技术中心）； National Taiwan University（国立台湾大学）

Comments Published at NeurIPS 2025. Project Page: https://chen-hao-chao.github.io/mdm-prime

2505.15754 2025-10-23 cs.LG cs.AI cs.RO

Improving planning and MBRL with temporally-extended actions

Palash Chatterjee, Roni Khardon

机构 * Indiana University（印第安纳大学）

Comments NeurIPS 2025. For project website, see https://pecey.github.io/MBRL-with-TEA/

2505.14489 2025-10-23 cs.AI cs.CL

Reasoning Models Better Express Their Confidence

Dongkeun Yoon, Seungone Kim, Sohee Yang, Sunkyoung Kim, Soyeon Kim, Yongil Kim, Eunbi Choi, Yireun Kim, Minjoon Seo

机构 * KAIST（韩国科学技术院）； LG AI Research（LG人工智能研究）； CMU（卡内基梅隆大学）； UCL（伦敦大学学院）

Comments Accepted to NeurIPS 2025

2505.13061 2025-10-23 cs.CV

3D Visual Illusion Depth Estimation

Chengtang Yao, Zhidan Liu, Jiaxi Zeng, Lidong Yu, Yuwei Wu, Yunde Jia

机构 * Beijing Key Laboratory of Intelligent Information Technology, School of Computer Science & Technology, Beijing Institute of Technology, China（北京智能信息科技重点实验室，计算机科学与技术学院，北京理工大学，中国）； Guangdong Provincial Key Laboratory of Machine Perception and Intelligent Computing, Shenzhen MSU-BIT University, Shenzhen, China（广东省机器感知与智能计算重点实验室，深圳MSU-BIT大学，深圳，中国）； NVIDIA

Comments NeurIPS 2025, Project: https://github.com/YaoChengTang/3D-Visual-Illusion-Depth-Estimation

2505.11883 2025-10-23 cs.LG cs.CV

MINGLE: Mixture of Null-Space Gated Low-Rank Experts for Test-Time Continual Model Merging

Zihuan Qiu, Yi Xu, Chiyuan He, Fanman Meng, Linfeng Xu, Qingbo Wu, Hongliang Li

机构 * University of Electronic Science and Technology of China（电子科技大学）； Dalian University of Technology（大连理工大学）

Comments accepted by NeurIPS 2025

2505.09160 2025-10-23 cs.LG cs.AI cs.IT cs.NI eess.SP math.IT

A Multi-Task Foundation Model for Wireless Channel Representation Using Contrastive and Masked Autoencoder Learning

Berkay Guler, Giovanni Geraci, Hamid Jafarkhani

机构 * Center for Pervasive Communications and Computing, University of California, Irvine（普及通信与计算中心，加州大学伊维特分校）； Nokia Standards（诺基亚标准）； Universitat Pompeu Fabra（庞培法布拉大学）

Comments - 17 pages, 7 figures, 5 tables - Submitted to IEEE JSAC Large AI Models for Future Wireless Communication Systems - Some of the results will appear in NeurIPS 2025, AI4NextG Workshop - This version is an extensive improvement in all aspects over the previous version with the same title - Dataset and implementation: https://github.com/BerkIGuler/WirelessContrastiveMaskedLearning

2503.21776 2025-10-23 cs.CV

Video-R1: Reinforcing Video Reasoning in MLLMs

Kaituo Feng, Kaixiong Gong, Bohao Li, Zonghao Guo, Yibing Wang, Tianshuo Peng, Junfei Wu, Xiaoying Zhang, Benyou Wang, Xiangyu Yue

机构 * CUHK MMLab（香港中文大学多模态实验室）； CUHK (SZ)（香港中文大学（深圳））； Tsinghua University（清华大学）； UCAS（中国科学院大学）； CUHK HCCL（香港中文大学高性能计算实验室）

Comments NeurIPS 2025, Project page: https://github.com/tulerfeng/Video-R1

2503.09271 2025-10-23 cs.CV cs.LG

DitHub: A Modular Framework for Incremental Open-Vocabulary Object Detection

Chiara Cappellino, Gianluca Mancusi, Matteo Mosconi, Angelo Porrello, Simone Calderara, Rita Cucchiara

机构 * AImageLab - University of Modena and Reggio Emilia（AImageLab - Modena和Reggio Emilia大学）

Comments Accepted at the 39th Conference on Neural Information Processing Systems (NeurIPS 2025)

2502.05625 2025-10-23 cs.LG

Training-Free Constrained Generation With Stable Diffusion Models

Stefano Zampini, Jacob K. Christopher, Luca Oneto, Davide Anguita, Ferdinando Fioretto

机构 * Polytechnic of Turin（都灵理工大学）； University of Virginia（弗吉尼亚大学）； University of Genoa（热那亚大学）

Comments Spotlight at the 39th Conference on Neural Information Processing Systems (NeurIPS 2025)

2412.00744 2025-10-23 cs.RO cs.AI

Open-World Drone Active Tracking with Goal-Centered Rewards

Haowei Sun, Jinwu Hu, Zhirui Zhang, Haoyuan Tian, Xinze Xie, Yufeng Wang, Xiaohua Xie, Yun Lin, Zhuliang Yu, Mingkui Tan

机构 * South China University of Technology（华南理工大学）； Institute for Super Robotics (Huangpu)（超机器人研究所）； Pazhou Laboratory（琶洲实验室）； Key Laboratory of Big Data and Intelligent Robot, Ministry of Education（大数据与智能机器人重点实验室）； Peng Cheng Laboratory（鹏城实验室）； Sun Yat-sen University（中山大学）； Harbin Engineering University（哈尔滨工程大学）

Comments NeurIPS 2025

2411.13022 2025-10-23 eess.IV cs.AI cs.CV cs.LG

Fast MRI for All: Bridging Access Gaps by Training without Raw Data

Yaşar Utku Alçalar, Merve Gülle, Mehmet Akçakaya

机构 * University of Minnesota（明尼苏达大学）

Comments Neural Information Processing Systems (NeurIPS), 2025 (Spotlight)

2411.03270 2025-10-23 cs.GT cs.LG

Stable Matching with Ties: Approximation Ratios and Learning

Shiyun Lin, Simon Mauras, Nadav Merlis, Vianney Perchet

机构 * INRIA, FairPlay Joint Team（INRIA，FairPlay联合团队）； Technion - Israel Institute of Technology（技术学院—以色列理工学院）

Comments Accepted to NeurIPS 2025