Real Time Anomaly Detection And Categorisation

TL;DR

This paper introduces a real-time anomaly detection and categorisation method that extends offline techniques to sequential data, enabling quick identification of point and collective anomalies with theoretical guarantees.

Contribution

It develops a novel online algorithm for anomaly detection and categorisation, with theoretical analysis and validation on simulated and real datasets.

Findings

Average run length to false alarm is comparable to offline methods

Detection delay is minimized and close to offline performance

Method effectively distinguishes between baseline, point, and collective anomalies

Abstract

The ability to quickly and accurately detect anomalous structure within data sequences is an inference challenge of growing importance. This work extends recently proposed post-hoc (offline) anomaly detection methodology to the sequential setting. The resultant procedure is capable of real-time analysis and categorisation between baseline and two forms of anomalous structure: point and collective anomalies. Various theoretical properties of the procedure are derived. These, together with an extensive simulation study, highlight that the average run length to false alarm and the average detection delay of the proposed online algorithm are very close to that of the offline version. Experiments on simulated and real data are provided to demonstrate the benefits of the proposed method.