Parameter selection in SVM with RBF kernel function

The E=l(sv)n method is proposed to be used to assess the performance of SVM instead of using exhaust algorithm, and it is obtained that using C γ=C~(constant) to define the optimization parameters will get good performance of