Using Mutual Information for Feature Selection in Network Intrusion Detection System

This paper presents the feature selection problem for data classification arising from a large number of redundant and irrelevant features. It first proposes a Mutual Information based Feature Selection Algorithm, MIFSA in short, that analytically selects the best features for classification. The key contribution is the use of mutual information, which can handle linearly and non-linearly dependent data features. Its effectiveness is evaluated in the cases of network intrusion detection. An Intrusion Detection System (IDS) is build using the feature selected by our proposed feature selection algorithm, named IDS+MIFSA. To verify the feasibility of IDS+MIFSA, several experiments are conducted on three well-known intrusion detection datasets: KDD Cup 99, NSL-KDD and Kyoto 2006+ dataset. The experimental results shows that our method performs better than other algorithms in most cases in terms of classification ccuracy.