Applied research in iatrology classification based on SVM

摘要

In the paper we introduced the soft margin SVC to solve linearly inseparable problems. Compared with the kernel trick, it is obvious that the two approaches actually solve the problems in different manners. Then we provided a novel view to design a kernel function based on a general proximity relation mapping. It shows better classification performance than the common Mercer kernels experimentally in the iatrology area.