Characterizing and Understanding Defense Methods for GNNs on GPUs

doi:10.1109/LCA.2023.3304638

	Characterizing and Understanding Defense Methods for GNNs on GPUs
	Wu, Meng 1,2; Yan, Mingyu 1; Yang, Xiaocheng 1; Li, Wenming 1; Zhang, Zhimin 1; Ye, Xiaochun 1; Fan, Dongrui 1,2
	2023-07-01
发表期刊	IEEE COMPUTER ARCHITECTURE LETTERS
ISSN	1556-6056
卷号	22 期号:2 页码:137-140
摘要	Graph neural networks (GNNs) are widely deployed in many vital fields, but suffer from adversarial attacks, which seriously compromise the security in these fields. Plenty of defense methods have been proposed to mitigate the impact of these attacks, however, they have introduced extra time-consuming stages into the execution of GNNs. These extra stages need to be accelerated because the end-to-end acceleration is essential for GNNs to achieve fast development and deployment. To disclose the performance bottlenecks, execution patterns, execution semantics, and overheads of the defense methods for GNNs, we characterize and explore these extra stages on GPUs. Given the characterization and exploration, we provide several useful guidelines for both software and hardware optimizations to accelerate the defense methods for GNNs.
关键词	Kernel Purification Estimation Graphics processing units Perturbation methods Electric breakdown Training Graph neural networks defense execution semantic execution pattern overhead
DOI	10.1109/LCA.2023.3304638
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program[2022YFB4501404] ; National Natural Science Foundation of China[62202451] ; CAS Project for Young Scientists in Basic Research[YSBR-029] ; CAS Project for Youth Innovation Promotion Association
WOS研究方向	Computer Science
WOS类目	Computer Science, Hardware & Architecture
WOS记录号	WOS:001082420100001
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/2XEOYT63/21118
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Yan, Mingyu
作者单位	1.Chinese Acad Sci, Inst Comp Technol, SKLP, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Wu, Meng,Yan, Mingyu,Yang, Xiaocheng,et al. Characterizing and Understanding Defense Methods for GNNs on GPUs[J]. IEEE COMPUTER ARCHITECTURE LETTERS,2023,22(2):137-140.
APA	Wu, Meng.,Yan, Mingyu.,Yang, Xiaocheng.,Li, Wenming.,Zhang, Zhimin.,...&Fan, Dongrui.(2023).Characterizing and Understanding Defense Methods for GNNs on GPUs.IEEE COMPUTER ARCHITECTURE LETTERS,22(2),137-140.
MLA	Wu, Meng,et al."Characterizing and Understanding Defense Methods for GNNs on GPUs".IEEE COMPUTER ARCHITECTURE LETTERS 22.2(2023):137-140.