Unsupervised intrusion detection algorithm based on association amendment

Abstract

Unsupervised fuzzy c-means clustering (FCM) algorithm is applied to intrusion detection so that intrusion detection system can directly deal with unlabeled original network data. Because particle swarm optimization (PSO) algorithm is easy to implement global optimum, FCM algorithm is improved based on particle swarm algorithm, in order to address the deficiencies that FCM is easy to fall into local optimum when applied to intrusion detection system. The unsupervised clustering result is further association amended and the accuracy and adaption of the intrusion detection system is improved.