Sparse Bayesian variable selection in kernel probit model for analyzing high-dimensional data

In this paper, we developed a sparse Bayesian variable selection in kernel probit model for high-dimensional data classification. Particularly we assigned a correlation prior distribution on the model size and a sparse prior distribution on the regression parameters. MCMC-based computation algorithms are outlined to generate samples from the posterior distributions. Simulation and real data studies show that in terms of the accuracy of variable selection and classification, our proposed method performs better than the other five Bayesian methods without the correlation term in the prior or those involving only one shrinkage parameter.