A Bayesian approach to support vector machines for the binary classification

The model of support vector machine (SVM) has been widely used to solve the problems of regression/classification. Here we propose a Bayesian approach to determining the separating hyperplane of an SVM, once its maximal margin is determined in the traditional way. This novel method minimizes the Bayes error in some derived direction. In the proposed model of -SVM, all the parameters are estimated by the reversible jump Markov chain Monte Carlo (RJMCMC) strategies, and the location parameter of decision boundary is finally described by a posterior distribution. Tested by many independent random experiments of 2-fold cross validations, the experimental results on some high-throughput biodata sets demonstrate the promising performance and robustness of this novel classification method.