Combined Kernel SVM and Its Application on Network Security Risk Evaluation
Cong‐Cong Li, Guo Ai-ling, Dan Li · 2008
Support vector machine SVM is a branch of artificial intelligence. SVM has many advantages in solving small sample size, nonlinear and high dimensional pattern recognition problem. Kernel function is the key technology of SVM, the choice of Kernel function will affect the learning ability and generalization ability of SVM, and different kernel function will construct different SVMS. At present, there are two types of kernel function, local kernel function which has better learning ability and whole kernel function which has better extensive ability. Since every traditional kernel function has its advantages and disadvantages, this paper analyze the principle of traditional kernel function and adopt a new kernel function of combined two kernel function, which called combined kernel function. It has better generalization ability and better learning ability, and adopt the combined kernel SVM into network security risk evaluation, compared with the SVM using traditional kernel. The result shows that the SVM based on combined kernels advance the speed of classification and has better classification precision than that with traditional kernels. The superiority and validity of this method is approved through experiment.