Detecting Anomalous Process Behaviour using Second Generation Artificial Immune Systems

Abstract. Artificial immune systems (AISs) are problem-solving systems inspired by the biological immune system. They have been successfully applied to a number of problem domains including fault tolerance, data mining and computer security. However, several algorithms central to many AISs have been shown to scale poorly and have low detection rates. However, the biological immune system is a very effective anomaly detector and we should be able to build AISs which do the same. AIS algorithms to date have largely been inspired by the adaptive immune system and by biologically-naive models. Our research is focussed on building more biologically-realistic AISs which are inspired by both the innate and adaptive immune systems. In this paper we present one such AIS and evaluate its performance on a realistic process anomaly detection problem. We show that this AIS performs better than standard AIS and policy-based anomaly detection methods, and is comparable in detection capability to stat...

• Type: Journal
• Year: 2010
• Where: CORR
• Authors: Jamie Twycross, Uwe Aickelin, Amanda M. Whitbrook
• Keywords: Anomaly Detection, Artificial Immune Systems, CORR 2010, Education, Immune Systems