Nonlinear system modeling based on KPCA and MKSVM

Abstract

Kernels are employed in Support Vector Machines (SVM) to map the nonlinear model into a higher dimensional feature space where the linear learning is adopted. Every kernel has its advantages and disadvantages. Preferably, the ‘good’ characteristics of two or more kernels should be combined. In this paper, the mathematical formulation of multiple kernel learning is given. To enhance the robust regression of the algorithm, KPCA is used for the support vectors' reduced process. Through the implementation for average molecular weight in polyacrylonitrile productive process, it demonstrates the good performance of the proposed method compared to single kernel.