结合经验贝叶斯的极大似然方法

A MAXIMUM LIKELIHOOD METHOD COMBINED WITH EMPRICAL BAYESIAN APPROACH

极大似然方法和贝叶斯方法是两种截然不同的统计方法 ,各有优缺点 .由于统计思想的不兼容 ,一般不结合起来使用 .本文根据统计学习理论的思想提出一种将两者结合使用的新方法 ,可以综合两者的优点 。

Maximum likelihood method and Bayesian approach are completely different statistical means.Both have their own advantages and disadvantages. However, because of their incompatibility in statistical thoughts,they are seldom used together. In this paper, a new method is proposed that can cornbine both of them according to statistical learning theory. It giyes a new solution to the multiple change points of variance problem.