先验概率未知时的多元Bayes判决Minimax方法  

Minimax method in multi-hypothesis Bayesian decision when priori probabilities are unknown

作　　者：何勇 朱允民[2] 

机构地区：[1]重庆科技学院数理系,重庆401331 [2]四川大学数学学院,成都610064

出　　处：《四川大学学报（自然科学版）》2009年第2期277-281,共5页Journal of Sichuan University(Natural Science Edition)

摘　　要：对于Bayes判决,需先知道各种假设的先验概率。当先验概率未知时,二元判决可用Minimax准则获得最坏先验概率时的Bayes判决。作者将这一结果推广到一般的多元Bayes判决中,给出了一般的多元Minimax方程组,并由此诱导出算法。数值例子显示本算法可行,For Bayesian decision, we must know priori probabilities. When the priori probabilities are unknown, the authors use minimax method to solve the binary decision. By extending this method to general multi-hypothesis decision, a group of minimax equations are derived, and corresponding algorithm to solve the minimax equations are presented. A numerical example shows the method and algorithm are efficient.

关 键 词：Minimax准则 Bayes准则 多元判决 Minimax方程组 

分 类 号：O211.9[理学—概率论与数理统计] O231.3[理学—数学]