Support Vector Machines (SVM’s) with various kernels have become very successful in pattern classification and regression. However, single kernels do not lead to optimal data models. Replacing the input space by a ...