Unsupervised Sequential Information Bottleneck Clustering For Building Anomaly Based Network Intrusion Detection Model.

Abstract

In this paper we present a novel approach to unsupervised clustering in building an efficient anomaly based network intrusion detection model. The method is based on a recently introduced sequential information bottleneck (sIB) principle. KDDCup 1999 intrusion detection benchmark dataset is used for the experimentation of our proposed technique. The experimental results demonstrate that the proposed technique is more suitable in detecting network intrusions in terms of accuracy compared to other existing clustering algorithms.