IDEA: Invariant defense for graph adversarial robustness

doi:10.1016/j.ins.2024.121171

	IDEA: Invariant defense for graph adversarial robustness
	Tao, Shuchang 1,2; Cao, Qi 1; Shen, Huawei 1,2; Wu, Yunfan 1,2; Xu, Bingbing 1; Cheng, Xueqi 1,2
	2024-10-01
发表期刊	INFORMATION SCIENCES
ISSN	0020-0255
卷号	680 页码:18
摘要	Despite the success of graph neural networks (GNNs), their vulnerability to adversarial attacks poses tremendous challenges for practical applications. Existing defense methods suffer from severe performance decline under unseen attacks, due to either limited observed adversarial examples or pre-defined heuristics. To address these limitations, we analyze the causalities in graph adversarial attacks and conclude that causal features are key to achieve graph adversarial robustness, owing to their determinedness for labels and invariance across attacks. To learn these causal features, we innovatively propose an Invariant causal DE fense method against adversarial Attacks (IDEA). We derive node-based and structure-based invariance objectives from an information-theoretic perspective. IDEA ensures strong predictability for labels and invariant predictability across attacks, which is provably a causally invariant defense across various attacks. Extensive experiments demonstrate that IDEA attains state-of-the-art defense performance under all five attacks on all five datasets. The implementation of IDEA is available at https:// github .com /TaoShuchang /IDEA _repo.
关键词	Invariant defense Adversarial robustness Causal feature Graph neural networks
DOI	10.1016/j.ins.2024.121171
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2022YFB3103700] ; National Key R&D Program of China[2022YFB3103701] ; Strategic Priority Research Program of the Chinese Academy of Sciences[XDB0680101] ; National Natural Science Foundation of China[62102402] ; National Natural Science Foundation of China[U21B2046] ; National Natural Science Foundation of China[62272125] ; Beijing Academy of Artificial Intelligence (BAAI)
WOS研究方向	Computer Science
WOS类目	Computer Science, Information Systems
WOS记录号	WOS:001302686700001
出版者	ELSEVIER SCIENCE INC
引用统计	被引频次：1[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/39616
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Cao, Qi; Shen, Huawei
作者单位	1.Chinese Acad Sci, Inst Comp Technol, CAS Key Lab AI Safety, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Tao, Shuchang,Cao, Qi,Shen, Huawei,et al. IDEA: Invariant defense for graph adversarial robustness[J]. INFORMATION SCIENCES,2024,680:18.
APA	Tao, Shuchang,Cao, Qi,Shen, Huawei,Wu, Yunfan,Xu, Bingbing,&Cheng, Xueqi.(2024).IDEA: Invariant defense for graph adversarial robustness.INFORMATION SCIENCES,680,18.
MLA	Tao, Shuchang,et al."IDEA: Invariant defense for graph adversarial robustness".INFORMATION SCIENCES 680(2024):18.