Cluster Based Ensemble Classification for Intrusion Detection System

Abstract

Network security is a challenging task, as there is a tremendous growth of network -based services and sharing of sensitive information on the network. Intrusion throws a serious risk in the network. Even though many hardening systems are developed against intrusions, conventional approaches like firewalls, virtual private networks, and encryption techniques are not enough to provide network security. Intrusion detection is a cyber security mechanism that plays an important role in securing the network. Detection rate and false alarm rate are the challenging issues to design an Intrusion Detection System (IDS). Various data mining techniques are used to implement network intrusion detection. Classification is a supervised learning which predicts the class label in the data set. Single classifier fails to obtain high accuracy. Base classifiers are not capable of detecting the attacks accurately. Ensemble classifier is a combination of base classifiers. Ensemble classifier outperforms base classifiers. In this paper, we propose a cluster- based ensemble classifier for IDS. K means clustering is used in the experiment. Ensemble classifier is built using ADTree and KNN. The experimental results show that the proposed ensemble classifier outperforms other classifiers with 99.8% accuracy.