Study on Bayesian Regression Model and Applications

Abstract

A sparse vector regression model is introduced. The algorithm is established by employing Gaussian process and Bayesian formulation. By using a special prior hyperparameter setting in the developing process, the number of parameters in the algorithm is reduced, and generating a relatively simple algorithm compared with similar type of Bayesian vector regression models. The algorithm is done by using computational iterative approach. The examples of applications to the function approximations and the inverse scattering problem are presented.