HYBRIDIZATION OF SUPPORT VECTOR MACHINE WITH CAT SWARM ALGORITHM FOR INTRUSION DETECTION

Abstract

Intrusion detection system (IDS) like firewall, access control and encryption mechanisms no longer provide the muchneeded security for systems and computer networks. Current IDS are developed on anomaly detection which helps in identifying attacks both known and unknown. Unfortunately, these anomaly-based IDS feature high false rate. In a bid to reduce this false alarm rate, this paper proposed an intrusion detection model based on Support Vector Machine (SVM) optimized with Cat swarm optimization (CSO) algorithm. Attribute reduction was carried out based on Information Gain (IG) and classification was performed based on the optimized Support vector. The result obtained shows that our model performs well with the least false alarm rate and good accuracy value compare with other classification algorithms evaluated using the same datasets.