Method of combining multi-class SVMs using dempster-shafer theory and its application

Abstract

To deal with multi-source multi-class problems, the method of combining multiple multi-class probability support vector machines (MPSVMs) using Dempster-Shafer evidence theory is proposed. The MPSVM is designed by mapping the outputs of standard support vector machines into a calibrated posterior probability using a learned sigmoid function, and then combining these learned probability support vector machines. The Dempster-Shafer evidence theory is used to combine these learned MPSVMs. Two schemes of combination are composed. One of the schemes takes into account the prior information. Our proposed method is applied to the fault diagnosis of a diesel engine. The experimental results show that the accuracy and robustness of fault diagnosis can be improved significantly.