Hybrid ensemble techniques used for classifier and feature selection in intrusion detection systems

The data security of networks is a universal problem for governments, companies, and persons. The frequency of internet attacks has grown substantially, as have attacker strategies. The solution to this problem is intrusion detection, a typical and successful methodology for planning intrusion detection systems (IDS) with machine learning. The proposed IDS method consists of three stages: pre-processing, feature selection, and classification. We remove duplicate data and normalised data in our method's first stage. Sequential forward floating selection (SFFS) with extra-tree use for feature selection removes unwanted features in our method's second stage. LogitBoost with extra-tree classification to use selected features in our method third stage. The proposed method is evaluated on standard datasets KDD CUP'99, NSL-KDD, UNSW-NB15, CICIDS2017, and CICIDS2018. The experimental results show that the proposed method outperforms the existing work in terms of accuracy, false alarm rate, and detection rate.