Ensemble learning for intrusion detection systems: A systematic mapping study and cross-benchmark evaluation

Intrusion detection systems (IDSs) are intrinsically linked to a comprehensive solution of cyberattacks prevention instruments. To achieve higher rate, the ability design an improved framework is sought after, particularly when utilizing ensemble learners. Designing often lies in two main challenges such as choice available base classifiers and combiner methods. This paper performs overview how learners exploited IDSs by means systematic mapping study. We collected analyzed 124 prominent publications from existing literature. The selected were then mapped into several categories years publications, publication venues, datasets used, methods, IDS techniques. Furthermore, this study reports analyzes empirical investigation new classifier approach, called stack (SoE) for anomaly-based IDS. SoE that adopts parallel architecture combine three individual random forest, gradient boosting machine, extreme machine homogeneous manner. performance significance among classification algorithms statistically examined terms their Matthews correlation coefficients, accuracies, false positive rates, area under ROC curve metrics. Our fills gap current literature concerning up-to-date study, not mention extensive evaluation recent advances learning techniques applied IDSs.