Interpolation of Scattered Data and Classifying in Support Vector Machine

摘要

The kernel function is important for support vector machines in classifying and regression. However, there are few papers to consult that how to select a kernel function for the given data. Since the effect of kernel mapping has not been understood very clearly, the result may be not as good as SVM should be in some case. In this paper, we present a method with interpolation based to construct a nonlinear transformation as one kind of kernel function according to the given data. The experiments show that we can find a hyperplane in our method, which has larger margin than that in canonical methods of SVM.