search for: svm_cross_valid

Hi, My question might be a little general. I have a number of values to select for the complexity parameters in some classifier, e.g. the C and gamma in SVM with RBF kernel. The selection is based on which values give the smallest cross validation error. I wonder if the randomized splitting of the available dataset into folds is done only once for all those choices for the parameter values, or