A novel algorithm for feature selection used in intrusion detection

Abstract

Intrusion detection systems play an important role in securing computer networks. The existing methods for intrusion detection deal with huge amount of data which contains irrelevant or redundant features. Accordingly, feature selection is critical for improving classification accuracy in an intrusion detection system. In this paper, we proposed a novel algorithm combining a variety of feature selection methods based on majority voting rule, and used the SVM as the basic classification algorithm. Experiments on NSL-KDD dataset indicate that the proposed algorithm selects superior feature subset than the state-of-the-art feature selection approaches used in the field of intrusion detection.