Presentation | 2021-09-09 Proposal of an Improving Method for the Laplacian Anomaly Detection of Temporal Networks Eriko Segawa, Toyoaki Taniguchi, Yusuke Sakumoto, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Many networks in the real world are dynamic and temporal wherein relationships among nodes change with time. Technologies to detect anomalies in dynamic networks are crucial to investigate various problems of society. The Laplacian anomaly detection (LAD) has been proposed as an innovative method for detecting anomalies in dynamic networks. LAD calculates the anomaly score, i.e., an indicator of the degree of deviation from the normal state, at each time from timeseries data for some eigenvalues of the Laplacian matrix, which represents the structure of networks. Using the calculated anomaly scores, LAD detects the anomaly in the dynamic network. The original LAD uses only the large eigenvalues of the Laplacian matrix; other eigenvalue combinations have not been discussed for anomaly detection. Based on spectral graph theory, small eigenvalues contain relevant information about the global structure of a network. Therefore, small eigenvalues are crucial for detecting large-scale anomalies in dynamic networks, and their use should improve the accuracy of LAD. Herein, we propose an improvement that uses not only the large eigenvalues but also the small eigenvalues of the Laplacian matrix for calculating anomaly scores in LAD. Through the evaluation, we clarify that the proposed improvement can significantly improve the accuracy of anomaly detection in LAD. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Anormaly Detection / Dynamic Network / Spectral Graph Theory / Laplacian Matrix / Social Network Analysis |
Paper # | CQ2021-40 |
Date of Issue | 2021-09-02 (CQ) |
Conference Information | |
Committee | CQ |
---|---|
Conference Date | 2021/9/9(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Wireless Communications Quality, 6G, IoT, Resource Management, Wireless Transmission, Cross layer Technologies, etc. |
Chair | Jun Okamoto(NTT) |
Vice Chair | Takefumi Hiraguri(Nippon Inst. of Tech.) / Gou Hasegawa(Tohoku Univ.) |
Secretary | Takefumi Hiraguri(NTT) / Gou Hasegawa(Ritsumeikan Univ.) |
Assistant | Yoshiaki Nishikawa(NEC) / Ryoichi Kataoka(KDDI Research) / Kimiko Kawashima(NTT) |
Paper Information | |
Registration To | Technical Committee on Communication Quality |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Proposal of an Improving Method for the Laplacian Anomaly Detection of Temporal Networks |
Sub Title (in English) | |
Keyword(1) | Anormaly Detection |
Keyword(2) | Dynamic Network |
Keyword(3) | Spectral Graph Theory |
Keyword(4) | Laplacian Matrix |
Keyword(5) | Social Network Analysis |
1st Author's Name | Eriko Segawa |
1st Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
2nd Author's Name | Toyoaki Taniguchi |
2nd Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
3rd Author's Name | Yusuke Sakumoto |
3rd Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
Date | 2021-09-09 |
Paper # | CQ2021-40 |
Volume (vol) | vol.121 |
Number (no) | CQ-173 |
Page | pp.pp.17-22(CQ), |
#Pages | 6 |
Date of Issue | 2021-09-02 (CQ) |