A Bayesian Meta-Modeling Approach for Gaussian Stochastic Process Models Using a Non Informative Prior
Hai-Song Deng, Yizhong Ma, Wenze Shao, Yiliu Tu · Communication in Statistics- Theory and Methods · 2012
In this article, an efficient Bayesian meta-modeling approach is proposed for Gaussian stochastic process models in computer experiments. Different prior densities and particularly, a non informative hyper prior have been employed on the parameters involved in the correlation matrix. And the estimation of related parameters is obtained by the expectation-maximization algorithm. Compared with the recent work of Li and Sudjianto (Citation2005), the proposed approach is not only of higher prediction accuracy but also of lower computational cost, due to the utilization of the non informative prior and the absence of tuning parameters. Experimental results demonstrate that our approach yields state-of-the-art performance.