arXivDaily arXiv每日学术速递 周一至周五更新

期刊&会议

ACM SIGKDD Conference on Knowledge Discovery and Data Mining · 会议 · Data Mining

2025-12-03 至 2025-12-03 共收录 3
2505.23337 2025-12-03 cs.LG cs.AI

Matryoshka Model Learning for Improved Elastic Student Models

用于改进弹性学生模型的Matryoshka模型学习

Chetan Verma, Aditya Srinivas Timmaraju, Cho-Jui Hsieh, Suyash Damle, Ngot Bui, Yang Zhang, Wen Chen, Xin Liu, Prateek Jain, Inderjit S Dhillon

机构 * Google(谷歌) Google DeepMind(谷歌DeepMind)

AI总结 本文提出MatTA框架,通过训练多个准确学生模型,利用TA模型提升模型性能,实现准确率与服务成本的平衡,实验显示在多个基准测试中取得显著提升。

Comments 10 pages, 5 figures, Accepted at KDD 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2506.12389 2025-12-03 cs.LG cs.AI stat.ML

Revisiting Clustering of Neural Bandits: Selective Reinitialization for Mitigating Loss of Plasticity

重新审视神经带客的聚类:用于缓解退化性的选择性重置

Zhiyuan Su, Sunhao Dai, Xiao Zhang

机构 * Gaoling School of Artificial Intelligence Renmin University of China Beijing China Gaoling School of Artificial Intelligence Renmin University of China

AI总结 本文提出SeRe框架,通过选择性重置缓解神经带客算法的退化性问题,提升其在动态环境中的适应性和鲁棒性。

Comments Accepted by KDD 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.05814 2025-12-03 cs.CR cs.CV cs.LG

Rank Matters: Understanding and Defending Model Inversion Attacks via Low-Rank Feature Filtering

秩至关重要:通过低秩特征过滤理解并防御模型反向攻击

Hongyao Yu, Yixiang Qiu, Hao Fang, Tianqu Zhuang, Bin Chen, Sijin Yu, Bin Wang, Shu-Tao Xia, Ke Xu

机构 * Tsinghua Shenzhen International Graduate School, Tsinghua University(清华大学深圳国际研究生院) Harbin Institute of Technology, Shenzhen(哈尔滨工业大学(深圳)) South China University of Technology(华南理工大学) Department of Computer Science and Technology, Tsinghua University(清华大学计算机科学与技术系)

AI总结 本文提出了一种基于低秩特征过滤的防御策略,通过减少中间表示的维度来有效防御模型反向攻击,实现了对多种攻击的有效防护。

Comments KDD 2026 Accept

详情

展开后加载摘要…

URL PDF HTML 收藏