Adaptive Adversarial Logits Pairing

doi:10.1145/3616375

CORC > 自动化研究所 > 中国科学院自动化研究所 > 多模态人工智能系统全国重点实验室

	Adaptive Adversarial Logits Pairing
	Wu, Shangxi 4; Sang, Jitao 3,4; Xu, Kaiyan 4; Zheng, Guanhua 2; Xu, Changsheng 1
刊名	ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
	2024-02-01
卷号	20 期号:2 页码:16
关键词	Adversarial defense adaptive dropout
ISSN号	1551-6857
DOI	10.1145/3616375
通讯作者	Wu, Shangxi(wushangxi@bjtu.edu.cn)
英文摘要	Adversarial examples provide an opportunity as well as impose a challenge for understanding image classification systems. Based on the analysis of the adversarial training solution-Adversarial Logits Pairing (ALP), we observed in this work that: (1) The inference of adversarially robust model tends to rely on fewer high-contribution features compared with vulnerable ones. (2) The training target of ALP does not fit well to a noticeable part of samples, where the logits pairing loss is overemphasized and obstructs minimizing the classification loss. Motivated by these observations, we design an Adaptive Adversarial Logits Pairing (AALP) solution by modifying the training process and training target of ALP. Specifically, AALP consists of an adaptive feature optimization module with Guided Dropout to systematically pursue fewer high-contribution features, and an adaptive sample weighting module by setting sample-specific training weights to balance between logits pairing loss and classification loss. The proposed AALP solution demonstrates superior defense performance on multiple datasets with extensive experiments.
资助项目	Fundamental Research Funds for the Central Universities[2023JBZY033] ; National Natural Science Foundation of China[61832002] ; National Natural Science Foundation of China[62172094] ; Beijing Natural Science Foundation[JQ20023] ; CCF-Zhipu AI Large Model Fund
WOS关键词	NEURAL-NETWORKS ; ROBUSTNESS
WOS研究方向	Computer Science
语种	英语
出版者	ASSOC COMPUTING MACHINERY
WOS记录号	WOS:001092595800026
资助机构	Fundamental Research Funds for the Central Universities ; National Natural Science Foundation of China ; Beijing Natural Science Foundation ; CCF-Zhipu AI Large Model Fund
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/54421]
专题	多模态人工智能系统全国重点实验室
通讯作者	Wu, Shangxi
作者单位	1.Chinese Acad Sci, Inst Automat, Beijing, Peoples R China 2.Univ Sci & Technol China, Beijing, Peoples R China 3.Tianjin Normal Univ, Tianjin, Peoples R China 4.Beijing Jiaotong Univ, Beijing Key Lab Traff Data Anal & Min, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Wu, Shangxi,Sang, Jitao,Xu, Kaiyan,et al. Adaptive Adversarial Logits Pairing[J]. ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS,2024,20(2):16.
APA	Wu, Shangxi,Sang, Jitao,Xu, Kaiyan,Zheng, Guanhua,&Xu, Changsheng.(2024).Adaptive Adversarial Logits Pairing.ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS,20(2),16.
MLA	Wu, Shangxi,et al."Adaptive Adversarial Logits Pairing".ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS 20.2(2024):16.