Validity Based Approach for Feature Selection in Intrusion Detection Systems

Abstract

Computer networks are vulnerable to growing number of security threats. The increase of attacks has caused obvious damages throughout the network at individual, enterprise, and government level. Intrusion detection systems are one of the tools that detect and remedy the presence of malicious activities. Intrusion detection systems face many challenges in terms of accurate analysis and evaluation. This paper proposes a new Intrusion detection system by deploying an entropy-based measure called V-measure to select significant features and reduce dimensionality while maintaining high accuracy in classification. The proposed intrusion detection system was tested on the CICIDS2017 dataset by applying machine learning classifiers such as Random Forest, Support Vector Machine, and RepTree algorithms. We then compared the results of the features selected with other features selection tools for correct classification of attacks. The expected results showed that the proposed method reduced irrelevant features and improved detection accuracy of the attacks while reducing the false positive rate.