Heterogeneous Ensemble Feature Selection for Network Intrusion Detection System

Intrusion detection systems get more attention to secure the computers and network systems. Researchers propose different network intrusion detection systems using machine learning techniques. However, the massive amount of data that contain irrelevant and redundant features is still challenging the intrusion detection systems. The redundancy and irrelevance of features may slow the processing time and decrease prediction performance. This paper proposes a Heterogeneous Ensemble Feature Selection (HEFS) method to select the relevant features while achieving better attack detection performance. The proposed method fuses the output feature subsets of five filter feature selection methods, using a union combination method, to obtain an ensemble features subset. HEFS method uses merit-based evaluation to avoid the internal redundancy of the obtained ensemble features subset and acquire the final optimal features. We evaluate the HEFS method with random forest, J48, random tree, and REP tree. In a multi-class NSL-KDD dataset, the experimental results show that the proposed method achieves better prediction performance than the specific feature selection methods and other frameworks.