Seminar on “Anomaly Detection in Temporal Networks through Topological Features and Motifs with Application to Mobility Data” at 2 pm on December 20, 2022

Speaker: Dr. Asim Kumer Dey, Brac University

Venue: ISRT seminar room

Date and time: 2 pm on Tuesday, 20 December 2022

Title: Anomaly Detection in Temporal Networks through Topological Features and Motifs with Application to Mobility Data

Abstract: This paper aims to shed light on the potential relations among various higher-order network topological features and meso-level functionality of temporal networks. In particular, we evaluate higher-order topological features, e.g., motif, betti number, and persistence images, to characterize the dynamics of a temporal network. We then use the functional data depth techniques on the resultant betti functions and vectorized persistent images to identify abnormal behavior in the temporal network. We apply the proposed methods to the temporal human mobility networks. Experiments on both synthetic temporal networks and human mobility networks demonstrate the effectiveness of the proposed methods.