Computer Network Security and Precaution Evaluation based on Incremental Relevance Vector Machine Algorithm and ACO

Abstract

Abstract It is well-known that the correct evaluation for computer network attack is very significant to avoid computer network attack. In order to improve the generalization performance of relevance vector machine, an incremental relevance vector machine algorithm(IRVM) based on ant colony optimization (ACO) is firstly presented and applied to computer network attack evaluation in the paper. In the experiments, we collect 85 samples of KDDCUP99 datasets as our experimental data, the detection accuracy of the hybrid algorithm of ACO and relevance vector machine is 97.65%, and the detection accuracy of the hybrid algorithm of relevance vector machine is 94.12%.It is shown that the detection accuracy of the hybrid algorithm of ACO and relevance vector machine for computer network attack is higher than that of relevance vector machine.