Influence to Nonlinear Classification Support Vector Machines by Separable Variables Kernel Function
变元可分离核函数对非线性支持向量分类机的影响

It is proved that under Hilbert space separable variables function can satisfy condition of Mercer theorem as kernel function to provide a new method in selecting new kernel function of nonlinear calssification machines based on support vector machine. Compared with other nonlinear classification support vector machines structured by known kernel function, the kernel function selected by new method can give better result.