PSO-based Semi-supervised Intrusion Detection Algorithm
Zhao Jian-fang · Journal of Chengdu University of Information Technology · 2012
View of unsupervised intrusion detection algorithm has low detection rate,supervised intrusion detection can not detect unknown attacker,this paper proposed a PSO-based semi-supervised Intrusion detection algorithm,it will expand a small amount of constraint information with density expansion method to obtain the clustering model,which used to guide the unlabeled data clustering,at last the unmarked data which was not determined its category,which using particle swarm optimization k-means algorithm to cluster to achieve the detection of abnormal.Algorithm improved the detection rate to 83.7% and the false positive rate decreased to 3.13%,the overall effect is superior to intrusion detection based on unsupervised and supervised learning.