Testing Detector Parameterization Using Evolutionary Exploit Generation

The testing of anomaly detectors is considered from the perspective of a Multi-objective Evolutionary Exploit Generator (EEG). Such a framework provides users of anomaly detection systems two capabilities. Firstly, no knowledge of protected data structures need be assumed. Secondly, the evolved exploits are then able to demonstrate weaknesses in the ensuing detector parameterization. In this work we focus on the parameterization of the second generation anomaly detector ‘pH’ and demonstrate how use of an EEG may identify weak parameterization of the detector.