AUCPro: AUC-Oriented Provable Robustness Learning

doi:10.1109/TPAMI.2025.3545639

	AUCPro: AUC-Oriented Provable Robustness Learning
	Bao, Shilong 1; Xu, Qianqian 2; Yang, Zhiyong 1; He, Yuan 3; Cao, Xiaochun 4; Huang, Qingming 1,2
	2025-06-01
发表期刊	IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
ISSN	0162-8828
卷号	47 期号:6 页码:4579-4596
摘要	The current studies of provable robustness for deep neural networks (DNNs) usually assume that the class distribution is overall balanced. However, in real-world applications especially for safety-sensitive systems, the class distribution often exhibits a long-tailed property. It is well-known that the Area Under the ROC Curve (AUC) is a more proper metric for long-tailed learning problems. Motivated by this fact, an AUC-oriented provable robustness learning framework (named AUCPro) is first proposed in this paper. The key is to construct a proxy model smoothed by the isotropic Gaussian noise and then consider optimizing the proxy model from the AUC-oriented learning point of view. Theoretically, we provide a certified safety region for AUCPro within which the model would be free from the & ell;(2 )adversarial attacks. Most importantly, we propose a novel standard to theoretically study the robustness generalization toward unseen data for provable robustness learning approaches. To the best of our knowledge, such a problem remains barely considered in the machine learning community. To be specific, under a general principle for performance-robustness trade-off, we prove that the generalization ability of the resulting model could be equivalently expressed as the expected adversarial risk of AUC under & ell;(2) perturbation. On top of this, we present two practical settings to explore the excess risk formed by the difference between the empirical risk of AUCPro and the derived generalization performance. Finally, comprehensive experiments speak to the efficacy of our proposed algorithm.
关键词	Robustness Training Perturbation methods Machine learning Heavily-tailed distribution Smoothing methods Gaussian noise Data mining Standards Protocols AUC-oriented learning adversarial robustness machine learning
DOI	10.1109/TPAMI.2025.3545639
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2018AAA0102000] ; National Natural Science Foundation of China[62441232] ; National Natural Science Foundation of China[62236008] ; National Natural Science Foundation of China[62025604] ; National Natural Science Foundation of China[62411540034] ; National Natural Science Foundation of China[U21B2038] ; National Natural Science Foundation of China[U23B2051] ; National Natural Science Foundation of China[62122075] ; National Natural Science Foundation of China[62206264] ; National Natural Science Foundation of China[92370102] ; Youth Innovation Promotion Association CAS ; Strategic Priority Research Program of the Chinese Academy of Sciences[XDB0680201] ; Postdoctoral Fellowship Program of CPSF[GZB20240729]
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:001484716600038
出版者	IEEE COMPUTER SOC
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/42378
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Xu, Qianqian; Huang, Qingming
作者单位	1.Univ Chinese Acad Sci, Sch Comp Sci & Technol, Beijing 101408, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100190, Peoples R China 3.Alibaba Grp, Secur Dept, Hangzhou 311121, Peoples R China 4.Sun Yat Sen Univ, Sch Cyber Sci & Technol, Shenzhen Campus, Shenzhen 518107, Peoples R China
推荐引用方式 GB/T 7714	Bao, Shilong,Xu, Qianqian,Yang, Zhiyong,et al. AUCPro: AUC-Oriented Provable Robustness Learning[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2025,47(6):4579-4596.
APA	Bao, Shilong,Xu, Qianqian,Yang, Zhiyong,He, Yuan,Cao, Xiaochun,&Huang, Qingming.(2025).AUCPro: AUC-Oriented Provable Robustness Learning.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,47(6),4579-4596.
MLA	Bao, Shilong,et al."AUCPro: AUC-Oriented Provable Robustness Learning".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 47.6(2025):4579-4596.