EagleMine: Vision-guided Micro-clusters recognition and collective anomaly detection

doi:10.1016/j.future.2020.08.033

	EagleMine: Vision-guided Micro-clusters recognition and collective anomaly detection
	Feng, Wenjie 1,2; Liu, Shenghua 1,2; Faloutsos, Christos 3; Hooi, Bryan 4; Shen, Huawei 1,2; Cheng, Xueqi 1,2
	2021-02-01
发表期刊	FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
ISSN	0167-739X
卷号	115 页码:236-250
摘要	Given a graph with millions of nodes, what patterns exist in the distributions of node characteristics? How can we detect them and separate anomalous nodes in a way similar to human visual perception? More generally, how can we identify micro-clusters in a histogram and spot some interesting patterns? In this paper, we propose a vision-guided algorithm, EagleMine, to recognize and summarize node groups in a histogram constructed from some correlated features. EagleMine hierarchically discovers node groups, which form internally connected dense areas in the histogram, by utilizing a water -level tree with multiple resolutions according to the rule of the visual recognition. EagleMine uses the statistical hypothesis test to determine the optimal groups while exploring the tree and simultaneously performs vocabulary-based summarization. Moreover, EagleMine can identify anomalous micro-clusters, consisting of nodes that exhibit very similar and suspicious behavior, deviate away from the majority. Experiments on the real-world datasets show that our method can recognize intuitive node groups as human vision does; it achieves the best summarization performance compared to baselines. In terms of anomaly detection, EagleMine also outperforms the state-of-the-art graph-based methods with significantly improving accuracy in a micro-blog dataset. Moreover, EagleMine can be used for other applications, e.g., to detect the synchronized patterns in the temporal retweet event. (c) 2020 Elsevier B.V. All rights reserved.
关键词	Pattern recognition Large graph mining Micro-clusters Anomaly detection Histogram
DOI	10.1016/j.future.2020.08.033
收录类别	SCI
语种	英语
资助项目	Strategic Priority Research Program of CAS, China[XDA19020400] ; NSF of China[61772498] ; NSF of China[61425016] ; NSF of China[91746301] ; NSF of China[61872206] ; Beijing NSF, China[4172059]
WOS研究方向	Computer Science
WOS类目	Computer Science, Theory & Methods
WOS记录号	WOS:000591438600010
出版者	ELSEVIER
引用统计	被引频次：6[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/2XEOYT63/16494
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Feng, Wenjie; Liu, Shenghua
作者单位	1.Chinese Acad Sci, Inst Comp Technol, CAS Key Lab Network Data Sci & Technol, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci UCAS, Beijing 100049, Peoples R China 3.Carnegie Mellon Univ, Comp Sci Dept, Pittsburgh, PA 15213 USA 4.Natl Univ Singapore, Dept Comp Sci, Singapore, Singapore
推荐引用方式 GB/T 7714	Feng, Wenjie,Liu, Shenghua,Faloutsos, Christos,et al. EagleMine: Vision-guided Micro-clusters recognition and collective anomaly detection[J]. FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE,2021,115:236-250.
APA	Feng, Wenjie,Liu, Shenghua,Faloutsos, Christos,Hooi, Bryan,Shen, Huawei,&Cheng, Xueqi.(2021).EagleMine: Vision-guided Micro-clusters recognition and collective anomaly detection.FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE,115,236-250.
MLA	Feng, Wenjie,et al."EagleMine: Vision-guided Micro-clusters recognition and collective anomaly detection".FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE 115(2021):236-250.