A Novel Multi-classification Intrusion Detection Model Based on Relevance Vector Machine

In view of the problems in the theory of support vector machine (SVM) and intrusion detection model, a new method of multi-classification intrusion detection model based on relevance vector machine (RVM) is proposed. Because RVM is based on Bayesian framework, a priori knowledge of the penalty term is introduced. The RVM algorithm needs less relevance vectors (RVs) (support vectors (SVs) in SVM) and it has better generalization ability than SVM. In order to get better classifier in anomaly detection, we analyze and model RVM algorithm using KDD99 dataset. Firstly, the Principal Components Analysis (PCA) is used to reduce the dimensionality of the feature vectors to enable better analysis of the data. Secondly, a multi-classification intrusion detection model based on relevance vector machine is designed to match these features. Finally, the matching forecast results of this model are achieved. The experiments show that this model has higher detection rate and better computational efficiency.