Intelligent hybrid anomaly network intrusion detection system

Abstract

Intrusion detection systems (IDSs) is an essential key for network defense. The hybrid intrusion detection system combines the individual base classifiers and feature selection algorithm to maximize detection accuracy and minimize computational complexity. We investigated the performance of Genetic algorithm-based feature selection system to reduce the data features space and then the hidden naïve bays (HNB) system were adapted to classify the network intrusion into five outcomes: normal, and four anomaly types including denial of service, user-to-root, remote-to-local, and probing. In order to evaluate the performance of introduced hybrid intrusion system, several groups of experiments are conducted and demonstrated on NSL-KDD dataset. Moreover, the performances of intelligent hybrid intrusion system have been compared with the results of well-known feature selection algorithms. It is found that, hybrid intrusion system produces consistently better performances on selecting the subsets of features which resulting better classification accuracies (98.63%). © 2011 Springer-Verlag.