Towards feature subset selection in intrusion detection

Abstract

Intrusion is a serious issue in computer and network systems because a single intrusion can cause a heavy loss in few seconds. To prevent an intrusion, a robust intrusion detection system is highly needed. Existing intrusion detection techniques are not robust; the number of false alarms is high. One of the reasons of false alarms is due to the use of a raw dataset that includes redundancy. To overcome this issue, the recent approaches used (PCA) for feature subset selection where features are first transformed into an eigen space and then features are selected based on their variances (i.e. eigenvalues), but the features corresponding to the highest eigenvalues may not have the optimal sensitivity for the classifier. Instead of using traditional approach of selecting features with the highest eigenvalues, an optimization approach is needed because the selection of most discriminative subset of transformed features is an optimization problem. One research used genetic algorithm (GA) to search the most discriminative subset of transformed features which is evolutionary optimization approach. The particle swarm optimization (PSO) is another optimization approach based on the behavioral study of animals/birds that outperforms GA in some applications. Therefore, the PSO based method is proposed in feature subset selection in this research work.