Improving fraud detection via hierarchical attention-based Graph Neural Network

doi:10.1016/j.jisa.2022.103399

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

	Improving fraud detection via hierarchical attention-based Graph Neural Network
	Liu, Yajing 1; Sun, Zhengya 2; Zhang, Wensheng 2
刊名	JOURNAL OF INFORMATION SECURITY AND APPLICATIONS
	2023-02-01
卷号	72 页码:10
关键词	Graph Neural Networks Fraud detection Attention mechanism
ISSN号	2214-2126
DOI	10.1016/j.jisa.2022.103399
通讯作者	Sun, Zhengya(zhengya.sun@ia.ac.cn)
英文摘要	Fraud has seriously influenced the social media ecosystems, and malicious users pursue high profit by disseminating fake information. Graph neural networks (GNN) have shown a promising potential for fraud detection tasks, where fraudulent nodes are identified by aggregating the neighbors that share similar feedbacks and relations. However, crafty fraudsters can trivially get around such detection via seemingly legitimate feedbacks once connected to legitimate users. In this paper, we leverage Relational Density Theory and propose a Hierarchical Attention-based Graph Neural Network (HA-GNN) for fraud detection, which incorporates weighted adjacency matrices across different relations against camouflage. This is motivated by the fact that there are dense connections between fraudsters who collectively participate in fraud activities. Specifically, we design a relation attention module to reflect the tie strength between two nodes, while a neighborhood attention module to capture the long-range structural affinity associated with the graph. We generate node embeddings by aggregating information from local/long-range structures and original node features. Experiments on three real-world datasets demonstrate that our approach achieves 3.21 - 9.97% RUC improvement compared with the state-of-the-arts.
资助项目	National Key R&D Pro-gram of China ; National Natural Science Foundation of China ; Natural Science Founda-tion of Beijing Municipality ; [2017YFC0803700] ; [61876183] ; [4172063]
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER
WOS记录号	WOS:000909640200001
资助机构	National Key R&D Pro-gram of China ; National Natural Science Foundation of China ; Natural Science Founda-tion of Beijing Municipality
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/51117]
专题	多模态人工智能系统全国重点实验室
通讯作者	Sun, Zhengya
作者单位	1.Univ Chinese Acad Sci, Beijing, Peoples R China 2.Chinese Acad Sci, Inst Automat, Res Ctr Precis Sensing & Control, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Liu, Yajing,Sun, Zhengya,Zhang, Wensheng. Improving fraud detection via hierarchical attention-based Graph Neural Network[J]. JOURNAL OF INFORMATION SECURITY AND APPLICATIONS,2023,72:10.
APA	Liu, Yajing,Sun, Zhengya,&Zhang, Wensheng.(2023).Improving fraud detection via hierarchical attention-based Graph Neural Network.JOURNAL OF INFORMATION SECURITY AND APPLICATIONS,72,10.
MLA	Liu, Yajing,et al."Improving fraud detection via hierarchical attention-based Graph Neural Network".JOURNAL OF INFORMATION SECURITY AND APPLICATIONS 72(2023):10.