Intrusion Detection Based on Support Vector Machine Divided up by Clusters

In order to solve the problem that algorithm SVM (Support Vector Machine) is very slowly for intrusion detection systems, a novel algorithm based on SVM divided up by clusters was proposed. In the method, Training set is divided into many subsets by clustering algorithm, and these subsets are classified by the decision function SVM. Detection Experiments with the algorithm on intrusion detection data were completed, the results show that the method can find the intrusion actions quickly with a high precision.