A Simplified Algorithm for Identifying Abnormal Changes in Dynamic Networks

TL;DR

This paper introduces a simplified, efficient algorithm for detecting abnormal changes in dynamic networks, maintaining performance while enabling early warning signals and local information highlighting.

Contribution

The paper presents a simplified version of an existing algorithm for abnormal change detection in dynamic networks, reducing complexity without sacrificing effectiveness.

Findings

The new algorithm maintains detection performance.

It effectively highlights local abnormal changes.

It can provide early warning signals in dynamic networks.

Abstract

Topological data analysis has recently been applied to the study of dynamic networks. In this context, an algorithm was introduced and helps, among other things, to detect early warning signals of abnormal changes in the dynamic network under study. However, the complexity of this algorithm increases significantly once the database studied grows. In this paper, we propose a simplification of the algorithm without affecting its performance. We give various applications and simulations of the new algorithm on some weighted networks. The obtained results show clearly the efficiency of the introduced approach. Moreover, in some cases, the proposed algorithm makes it possible to highlight local information and sometimes early warning signals of local abnormal changes.