Heterogeneous Ensemble Feature Selection for Network Intrusion Detection System

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