arXivDaily arXiv每日学术速递 周一至周五更新

期刊&会议

NeurIPS

Conference on Neural Information Processing Systems · 会议 · Machine Learning

2026-05-13 至 2026-05-13 共收录 5
2508.20614 2026-05-13 stat.ML cs.LG stat.CO

Improving the Accuracy of Amortized Model Comparison with Self-Consistency

通过自一致性提升近似模型比较的准确性

Šimon Kucharský, Aayush Mishra, Daniel Habermann, Stefan T. Radev, Paul-Christian Bürkner

机构 * Department of Statistics TU Dortmund University(统计系杜伊斯堡-艾森大学) Department of Cognitive Science Rensselaer Polytechnic Institute(认知科学系拉特格斯理工学院)

AI总结 本文评估了四种近似模型比较方法,并通过自一致性损失提升在分布偏移下的性能。在封闭世界场景中,分类器表现良好,但在开放世界场景中,自一致性训练显著提升了模型比较估计。

Comments 22 pages, 14 figures. This version extends our initial results presented at Reliable ML from Unreliable Data Workshop at NeurIPS 2025. Previously, this version appeared as arXiv:2512.14308v2, which has now been withdrawn: the two versions share too much content to be considered separate papers

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.20535 2026-05-13 cs.LG

Rotary Masked Autoencoders are Versatile Learners

旋转掩码自编码器是通用学习者

Uros Zivanovic, Serafina Di Gioia, Andre Scaffidi, Martín de los Rios, Gabriella Contardo, Roberto Trotta

机构 * University of Trieste(特里埃斯特大学) Abdus Salam International Centre for Theoretical Physics (ICTP)(阿布杜斯·萨拉姆国际理论物理学中心(ICTP)) Scuola Internazionale Superiore di Studi Avanzati (SISSA)(国际先进研究高等学院(SISSA)) University of Nova Gorica(诺瓦戈里察大学) INFN – National Institute for Nuclear Physics(意大利国家核物理研究所(INFN)) ICSC - Centro Nazionale di Ricerca in High Performance Computing(高性能计算国家研究中心(ICSC)) Imperial College London(伦敦帝国理工学院)

AI总结 RoMAE通过旋转位置嵌入实现多维连续位置学习,无需时间序列特化架构,在多种模态上表现优异,超越专门时间序列架构。

Comments NeurIPS 2025 Final Camera Ready

Journal ref Advances in Neural Information Processing Systems 38, NeurIPS 2025, Pages 133952-133987

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.11398 2026-05-13 cs.AI cs.CL

AcuityBench: Evaluating Clinical Acuity Identification and Uncertainty Alignment

AcuityBench:评估语言模型对医疗紧急情况识别和不确定性对齐

Robin Linzmayer, Georgianna Lin, Di Coneybeare, Jason Chu, Trudi Cloyd, Manish Garg, Miles Gordon, Elizabeth Hartofilis, Benjamin Hong, Ashraf Hussain, Eugene Y. Kim, Oluchi Iheagwara King, Ross McCormack, Erica Olsen, John K. Riggins, Mustafa N. Rasheed, Dana L. Sacco, Vinay Saggar, Osman R. Sayan, Amit Shembekar, Janice Shin-Kim, Wendy W. Sun, Bernard P. Chang, David Kessler, Noémie Elhadad

机构 * Department of Computer Science, Columbia University, New York, NY, USA(计算机科学系,哥伦比亚大学,纽约,纽约州,美国) Department of Biomedical Informatics, Columbia University, New York, NY, USA(生物医学信息学系,哥伦比亚大学,纽约,纽约州,美国) Department of Emergency Medicine, Columbia University Irving Medical Center, New York, NY, USA(急诊医学系,哥伦比亚大学伊文思医疗中心,纽约,纽约州,美国)

AI总结 AcuityBench通过统一框架评估语言模型对医疗紧急程度的识别能力,包含914个案例,涵盖明确和模糊情况,揭示模型在不同任务格式下的表现差异及不确定性处理问题。

Comments 41 pages, 5 figures. Preprint under review for the Track on Evaluations and Datasets at NeurIPS 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.11316 2026-05-13 cs.LG math.OC

Error whitening: Why Gauss-Newton outperforms Newton

误差白化:为何高斯-牛顿法优于牛顿法

Maricela Best McKay, Nathan P. Lawrence, Brian Wetton, R. Bhushan Gopaluni

机构 * University of British Columbia(不列颠哥伦比亚大学) University of California, Berkeley(加州大学伯克利分校)

AI总结 本文从函数空间视角分析高斯-牛顿法优于牛顿法的原因,指出通过投影消除参数化扭曲,实现误差白化,从而在不同学习任务中表现更优。

Comments Neurips preprint

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.10985 2026-05-13 cs.LG cs.AI q-bio.BM

Structural Interpretations of Protein Language Model Representations via Differentiable Graph Partitioning

通过可微分图划分对蛋白质语言模型表示进行结构解释

Siddhant Dutta, Edward Tan Beng Wai, Soumick Sarker, Pasan Gunawardane, Jagath C. Rajapakse

机构 * Nanyang Technological University(南洋理工大学)

AI总结 本文提出一种插件式框架,通过将ESM-2表示投影到蛋白质接触图并应用SoftBlobGIN网络,实现结构感知的消息传递和学习功能子结构,提升下游任务性能。

Comments 19 Pages, 8 figures, 11 Tables, Submitted to NeurIPS 2026

详情

展开后加载摘要…

URL PDF HTML 收藏