您好，欢迎访问中国农业科学院 机构知识库！

全部

  • 全部
  • 标题
  • 学者
  • 机构
  • 关键词

Generalization analysis of adversarial pairwise learning

文献类型： 外文期刊

作者： Wen, Wen 1 ; Li, Han 1 ; Wu, Rui 5 ; Wu, Lingjuan 1 ; Chen, Hong 1 ;

作者机构： 1.Huazhong Agr Univ, Coll Informat, Wuhan 430070, Peoples R China

2.Minist Educ, Engn Res Ctr Intelligent Technol Agr, Wuhan 430070, Peoples R China

3.Huazhong Agr Univ, Shenzhen Inst Nutr & Hlth, Shenzhen 518000, Peoples R China

4.Chinese Acad Agr Sci, Shenzhen Branch, Guangdong Lab Lingnan Modern Agr, Genome Anal Lab,Minist Agr,Agr Genom Inst Shenzhen, Shenzhen 518000, Peoples R China

5.Horizon Robot, Beijing 100190, Peoples R China

关键词： Adversarial pairwise learning; Perturbation attacks; Error analysis; Generalization bounds

期刊名称：NEURAL NETWORKS （ 影响因子：6.3； 五年影响因子：7.5 ）

ISSN： 0893-6080

年卷期： 2025 年 183 卷

页码：

收录情况： SCI

摘要： Adversarial pairwise learning has become the predominant method to enhance the discrimination ability of models against adversarial attacks, achieving tremendous success in various application fields. Despite excellent empirical performance, adversarial robustness and generalization of adversarial pairwise learning remain poorly understood from the theoretical perspective. This paper moves towards this by establishing the high- probability generalization bounds. Our bounds generally apply to various models and pairwise learning tasks. We give application examples involving explicit bounds of adversarial bipartite ranking and adversarial metric learning to illustrate how the theoretical results can be extended. Furthermore, we develop the optimistic generalization bound at order c(n-1) on the sample size n by leveraging local Rademacher complexity. Our analysis provides meaningful theoretical guidance for improving adversarial robustness through feature size and regularization. Experimental results validate theoretical findings.

  • 相关文献
作者其他论文 更多>>
置顶
购物车
反馈

© 京ICP备09089781号-29 主办单位：中国农业科学院

学术资源： 中国农业科学院 农业专业知识服务系统 农科联盟资源共建共享平台 中国农业科技文献与信息服务平台 中国农业期刊集成服务平台