Exploring Ensemble-Based Class Imbalance Learners for Intrusion Detection in Industrial Control Networks

Classifier ensembles have been utilized in the industrial cybersecurity sector for many years. However, their efficacy and reliability intrusion detection systems remain questionable current research, owing to particularly imbalanced data issue. The purpose of this article is address a gap literature by illustrating benefits ensemble-based models identifying threats attacks cyber-physical power grid. We provide framework that compares nine cost-sensitive individual ensemble designed specifically handling data, including C4.5, roughly balanced bagging, random oversampling undersampling synthetic minority boosting, AdaC2, EasyEnsemble. Each ensemble’s performance tested against range benchmarked system datasets utilizing accuracy, Kappa statistics, AUC metrics. Our findings demonstrate EasyEnsemble outperformed significantly comparison its rivals across board. Furthermore, strategies were effective boosting-based but not bagging-based ensemble.