An Improved Model of Anomaly Detection Using Two-Level Classifier Ensemble

Network infrastructures are in jeopardy of suffering nowadays since a number of attacks have been developed and grown up enormously. In order to get rid of such security threats, a defense mechanism is much sought-after. This paper proposes an improved model of intrusion detection by using two-level classifier ensemble. The proposed model is made up of a PSO-based feature selection technique and a two-level classifier ensemble which employs two ensemble learners, i.e. boosting and random subspace model (RSM). The experiment conducted on NSL-KDD dataset reveals that the proposed model outperforms previous detection models significantly in terms of accuracy and false alarm rate (FPR).