Optimizing parameters of support vector machine based on gradient algorithm

Some recent methods of parameters' optimization are analyzed,and several characteristics of efficient arithmetics to optimize parameters of support vector machine are generalized.Gradient algorithm can not be utilized directly because of its requirement of differentiable function.In the new method proposed,gradient direction is not directional derivative but optimized result of chaos search in local area.This method doesn't require differentiable function,and has the advantage of faster convergent speed,the ability of optimization within global scope and the independence between eventual optimized paramaters and initial SVM paramaters.