Anomaly Discrimination and Categorization

Abstract

In Chap. 14 the issues of characterizing anomalies from various aspects are investigated. This chapter looks into the issues of how to discriminate and categorize anomalies. Despite that anomaly discrimination was also studied by Adaptive Causal Anomaly Detector (ACAD) in Chap. 14 anomaly discrimination presented in this chapter is quite different from ACAD in the sense that it does not require causality as well as building an anomaly library as ACAD does. It is known that anomaly detection finds data sample vectors whose signatures are spectrally distinct from their surrounding data sample vectors. Unfortunately, it generally cannot discriminate its detected anomalies one from another. One common approach is to measure closeness of spectral characteristics among detected anomalies to determine if the detected anomalies are actually targets of different types. However, this also leads to a challenging issue of how to find an appropriate criterion to threshold their spectral similarities. This chapter investigates the issue of anomaly discrimination without appealing for any spectral measure. The idea is to take advantage of an unsupervised target detection algorithm coupled with an anomaly detector to discriminate detected anomalies which can be further categorized into different types of targets.