The Value of Summary Statistics for Anomaly Detection in Temporally-Evolving Networks: A Performance Evaluation Study

TL;DR

This study evaluates the effectiveness of summary statistics in detecting anomalies in dynamic networks, emphasizing the importance of temporal auto-correlation and providing insights into their practical utility.

Contribution

It offers a comprehensive evaluation of summary statistics for anomaly detection in evolving networks, incorporating temporal auto-correlation and various network conditions.

Findings

Summary statistics often outperform complex methods in anomaly detection.

Temporal auto-correlation significantly impacts detection performance.

Performance varies with anomaly type, duration, and network sparsity.

Abstract

Network data has emerged as an active research area in statistics. Much of the focus of ongoing research has been on static networks that represent a single snapshot or aggregated historical data unchanging over time. However, most networks result from temporally-evolving systems that exhibit intrinsic dynamic behavior. Monitoring such temporally-varying networks to detect anomalous changes has applications in both social and physical sciences. In this work, we perform an evaluation study of summary statistics for anomaly detection in temporally-evolving networks by incorporating principles from statistical process monitoring. In contrast to previous studies, we deliberately incorporate temporal auto-correlation in our study. Other considerations in our comprehensive assessment include types and duration of anomaly, model type, and sparsity in temporally-evolving networks. We conclude that summary statistics can be valuable tools for network monitoring and often perform better than more complicated statistics.