Intelligent Detection of Major Network Attacks Using Feature Selection Methods

Intrusion detection system (IDS) detects an illegal exploitation of computer systems. In intrusion detection systems, feature selection plays an important role in a sense of improving classification performance and reducing the computational complexity. In this paper, we focus on improving identification of major network attacks like DoS, R2L and Probe using various feature selection techniques (IG, CHI2 and OCFS). This research work explored the possibility of employing a variety of classifiers, but limited to J48, Naive Bayes and AdaBoost. Empirical evaluations were completed based on a standard network intrusion data set (KDDCUP99). The Experimental results show that the feature selection approach gives considerable increase of performance in detecting network intrusions as compared to normal approach.