Application Research of Support Vector Machine in Network Security Risk Evaluation

Along with the extensive application of the network, network security has received increasing attention recently.This paper researches on the network security risk evaluation and analyze the traditional risk evaluation methods, then proposes a new network security risk evaluation method based on Support Vector Machine (SVM) and Binary tree. Unlike the traditional risk evaluation methods, SVM is a novel type of learning machine technique which developed on structural risk minimization principle.SVM has many advantages in solving small sample size, nonlinear and high dimensional pattern recognition problem.The principles of SVM and binary tree are introduced in detail and apply it into network security risk assessment, it divided risk rate of network security into 4 different rates and more .Compare to ANN about the Classification precision, Generalization Performance, learning and testing time, it indicates that SVM has higher Classification precision, better generalization Performance and less learning and testing time, especially get a better assessment performance under small samples. It indicates that SVM has absolute superiority on network security risk evaluation, the validity and superiority of this method is approved through the experiment.