基于贝叶斯多层先验的元建模研究

Research on Meta-modeling Using Bayesian Hierarchical Priors

计算机试验成为一种研究复杂物理系统的替代方法,它引起了人们的关注。元模型的求解问题是计算机试验中研究其它问题的基础。通过贝叶斯多层先验,提出了一种新的计算机试验中求解元模型的方法。新方法通过EM(expectation-maximization)算法求解,并且将之与国际上的几种已有的方法进行了比较。试验结果表明,这种新的方法取得了较理想的试验结果,并且大大地较少了计算的负担。

Computer experiments have become an attractive alternative to study complex physical systems. Computer experiments are primarily on the task of recta-modeling, which is the central to achieving any goal in computer experiments. A novel approach of meta-modeling was proposed for computer experiments via Bayesian hierarchical prior. Implementation was carried out by an EM （expectationmaximization） algorithm and some other methods were compared with. Experimental results demonstrate that the new approach not only yields state-of-art performance, but also has low computational cost.