Evaluation of simple performance measures for tuning SVM hyperparameters

Choosing optimal hyperparameters for support vector machines is an important step in SVM design. This is usually done by minimizing either an estimate of generalization error or some other related performance measures. In this paper, we empirically study the usefulness of several simple performance measures that are very inexpensive to compute. The results point out which of these performance measures are adequate functionals for tuning SVM hyperparameters. For SVMs with L1 soft-margin formulation, none of the simple measures yields a performance as good as k-fold cross-validation.