Mitigating Semantic Collapse in Partially Relevant Video Retrieval
Comments Accpeted to NeurIPS 2025. Code is available at https://github.com/admins97/MSC_PRVR
期刊&会议
NeurIPS
Conference on Neural Information Processing Systems · 会议 · Machine Learning
Who Does Your Algorithm Fail? Investigating Age and Ethnic Bias in the MAMA-MIA Dataset
机构 * DTU Compute（丹麦技术大学计算学院）
Comments Medical Imaging Meets EurIPS (NeurIPS-endorsed workshop) - MedEurIPS
Soft Task-Aware Routing of Experts for Equivariant Representation Learning
机构 * Department of Statistics and Data Science, Yonsei University（统计与数据科学系，延世大学）
Comments NeurIPS 2025
GUI-Rise: Structured Reasoning and History Summarization for GUI Navigation
机构 * ShanghaiTech University（上海科技大学） ； ByteDance（字节跳动） ； Shanghai Engineering Research Center of Intelligent Vision and Imaging（上海智能视觉与成像工程研究中心）
Comments Published in NeurIPS 2025
Exploring Landscapes for Better Minima along Valleys
机构 * State Key Lab of Processors, Institute of Computing Technology, Chinese Academy of Sciences（中国科学院过程工程研究所状态关键实验室） ； University of Chinese Academy of Sciences（中国科学院大学）
Comments Neurips 2025 poster
ZEBRA: Towards Zero-Shot Cross-Subject Generalization for Universal Brain Visual Decoding
机构 * The Hong Kong University of Science and Technology（香港科学与技术大学）
Comments Accepted by NeurIPS 2025
Causal Masking on Spatial Data: An Information-Theoretic Case for Learning Spatial Datasets with Unimodal Language Models
机构 * Department of Electrical and Computer Engineering（电气与计算机工程系） ； Johns Hopkins University（约翰霍普金斯大学） ； Whiting School of Engineering（工程学院）
Comments 8 pages, NeurIPS 2025
Gradient Descent as Loss Landscape Navigation: a Normative Framework for Deriving Learning Rules
机构 * Harvard Medical School（哈佛医学院） ； Harvard University（哈佛大学）
Comments Accepted to NeurIPS 2025
MM-OPERA: Benchmarking Open-ended Association Reasoning for Large Vision-Language Models
机构 * Sun Yat-sen University（中山大学） ； Peng Cheng Laboratory（鹏城实验室） ； Jinan University（暨南大学）
Comments NeurIPS 2025 Datasets and Benchmarks Track poster
DC4GS: Directional Consistency-Driven Adaptive Density Control for 3D Gaussian Splatting
Comments Accepted to NeurIPS 2025 / Project page: https://github.com/cgskku/dc4gs
CAS-Spec: Cascade Adaptive Self-Speculative Decoding for On-the-Fly Lossless Inference Acceleration of LLMs
机构 * TeleAI ； Shanghai Jiao Tong University（上海交通大学） ； Hong Kong University of Science and Technology（香港科技大学）
Comments 10 pages, 3 figures, NeurIPS 2025 poster
'Studies for': A Human-AI Co-Creative Sound Artwork Using a Real-time Multi-channel Sound Generation Model
机构 * Sony Group Corporation（索尼集团）
Comments Accepted at NeurIPS Creative AI Track 2025, 9 pages, 6 figures, 1 table, Demo page: https://sony.github.io/studies-for/
Adaptive Stochastic Coefficients for Accelerating Diffusion Sampling
机构 * AGI Lab, Westlake University（西交大实验室，西湖大学） ； Nanyang Technological University（南洋理工大学） ； University of Chinese Academy of Sciences（中国科学院大学） ； Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所） ； Tongji University（同济大学）
Comments To appear in NeurIPS 2025
Hankel Singular Value Regularization for Highly Compressible State Space Models
机构 * Courant Institute of Mathematical Sciences（Courant数学科学研究所） ； New York University（纽约大学）
Comments Accepted at NeurIPS 2025
I-RAVEN-X: Benchmarking Generalization and Robustness of Analogical and Mathematical Reasoning in Large Language and Reasoning Models
机构 * IBM Research – Zurich（IBM瑞士研究中心） ； ETH Zurich（苏黎世联邦理工学院）
Comments Accepted at the 5th Workshop on Mathematical Reasoning and AI (MATH-AI), NeurIPS 2025
Dynamic Gaussian Splatting from Defocused and Motion-blurred Monocular Videos
机构 * Sun Yat-sen University（中山大学）
Comments Accepted to NeurIPS 2025
FastLongSpeech: Enhancing Large Speech-Language Models for Efficient Long-Speech Processing
机构 * Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Chinese Academy of Sciences (ICT/CAS)（智能信息处理重点实验室，计算技术研究所，中国科学院） ； Key Laboratory of AI Safety, Chinese Academy of Sciences（人工智能安全重点实验室，中国科学院） ； University of Chinese Academy of Sciences, Beijing, China（中国科学院大学，北京，中国） ； School of Future Science and Engineering, Soochow University（未来科学与工程学院，苏州大学）
Comments NeurIPS 2025. The code is at https://github.com/ictnlp/FastLongSpeech. This model is at https://huggingface.co/ICTNLP/FastLongSpeech. The dataset is at https://huggingface.co/datasets/ICTNLP/LongSpeech-Eval
Graph Diffusion that can Insert and Delete
Comments 39th Conference on Neural Information Processing Systems (NeurIPS 2025). Equation 2 is now correct
RADAR: Benchmarking Language Models on Imperfect Tabular Data
机构 * Google Research（谷歌研究） ； Google DeepMind（谷歌DeepMind） ； University of Washington（华盛顿大学）
Comments NeurIPS 2025 Dataset and Benchmark Track
Kernel conditional tests from learning-theoretic bounds
机构 * Institute for Data Science in Mechanical Engineering, RWTH Aachen University（机械工程数据科学研究所，亚琛工业大学） ； Department of Mathematics, School of Computation, Information and Technology, Technical University of Munich（数学系，计算、信息与技术学院，慕尼黑技术大学） ； Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心）
Comments 46 pages, 8 figures, 9 tables. Accepted at NeurIPS 2025; to appear in the proceedings of the Thirty-ninth Annual Conference on Neural Information Processing Systems. Reviews and discussion: https://openreview.net/forum?id=hJKDwf32Xu
AstroVisBench: A Code Benchmark for Scientific Computing and Visualization in Astronomy
机构 * The University of Texas at Austin（德克萨斯大学奥斯汀分校） ； NSF National Optical-Infrared Astronomy Research Laboratory（国家光学红外天文研究实验室） ； University of Virginia（弗吉尼亚大学） ； SLAC National Accelerator Laboratory（斯坦福直线加速器实验室） ； Texas Advanced Computing Center（德克萨斯高级计算中心）
Comments Accepted at NeurIPS 2025 Datasets & Benchmarks Track
R$^2$ec: Towards Large Recommender Models with Reasoning
机构 * The Hong Kong Polytechnic University（香港理工大学） ； National University of Singapore（新加坡国立大学） ； University of Science and Technology of China（中国科学技术大学） ； Harbin Institute of Technology (Shenzhen)（哈尔滨工业大学（深圳））
Comments Accepted by Neurips 2025
Scaling Diffusion Transformers Efficiently via $μ$P
机构 * Gaoling School of AI, Renmin University of China（中国人民大学人工智能学院） ； Beijing Key Laboratory of Research on Large Models（北京大型模型研究关键实验室） ； Engineering Research Center of Next-Generation Intelligent Search（下一代智能搜索与推荐工程研究中心） ； Tsinghua University（清华大学）
Comments Accepted by NeurIPS 2025, 38 pages, 10 figures, 17 tables
Reinforcing the Diffusion Chain of Lateral Thought with Diffusion Language Models
机构 * Zhejiang Univeristy（浙江大学） ； MAPLE Lab, Westlake University（西湖大学MAPLE实验室） ； Matterwave Intelligence ； Institute of Advanced Technology, Westlake Institute for Advanced Study（西湖研究所以高级技术研究所）
Comments Accepted to NeurIPS 2025. Code link: https://github.com/maple-research-lab/LLaDOU
NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation
机构 * National University of Singapore（新加坡国立大学） ； Sea AI Lab（Sea AI实验室）
Comments NeurIPS 2025
D$^2$USt3R: Enhancing 3D Reconstruction for Dynamic Scenes
机构 * KAIST AI（韩国国立科学技术院人工智能研究所） ； Sony AI（索尼人工智能实验室） ； ETH Zürich AI Center, CVG, PRS（苏黎世联邦理工学院人工智能中心、计算机视觉组、人机交互组） ； Sony Group Corporation（索尼集团）
Comments NeurIPS 2025; project page: https://cvlab-kaist.github.io/DDUSt3R/
More of the Same: Persistent Representational Harms Under Increased Representation
机构 * EleutherAI ； Universitat Ramon Llull, ESADE（拉蒙·拉鲁尔大学，ESADE） ； UT Austin（德克萨斯大学奥斯汀分校）
Comments Proceedings of the Neural Information Processing Systems (NeurIPS) 2025; 39 pages, 7 figures, 15 tables
A Regularized Newton Method for Nonconvex Optimization with Global and Local Complexity Guarantees
Comments NeurIPS 2025
DPA: A one-stop metric to measure bias amplification in classification datasets
机构 * Arizona State University（亚利桑那州立大学）
Comments Accepted at NeurIPS 2025
PROFIT: A Specialized Optimizer for Deep Fine Tuning
机构 * GM Cruise LLC
Comments technical report, 23 pages, NeurIPS 2025 poster
