贝叶斯学习中基于贝叶斯判别分析的先验分布选取  

Choosing a Suitable Prior for Bayesian Learning Based on Bayesian Discrimination

作　　者：胡振宇[1] 林士敏[1] 陆玉昌[2] 

机构地区：[1]广西师范大学计算机科学系,桂林541004 [2]清华大学计算机科学与技术系,北京100084

出　　处：《计算机科学》2003年第8期134-135,共2页Computer Science

基　　金：智能技术与系统国家重点实验室开放课题(99002)

摘　　要：In this paper we propose an experimental method to choose a prior distribution. Different from many re-searchers, who offered lots of principles that separated from sample information, we consider it a Bayesian discrimina-tion problem combining with the sample information. We introduce the concept of Posterior belief about prior distri-butions. With the well-known Bayes theorem we give out a formula to calculate it and propose a method to discrirni-nate a prior between prior distributions-- Highest Posterior Belief (HPB). We also show that under certain condition,the HPB method is identical with the ML-I method.In this paper we propose an experimental method to choose a prior distribution. Different from many researchers, who offered lots of principles that separated from sample information, we consider it a Bayesian discrimination problem combining with the sample information. We introduce the concept of Posterior belief about prior distributions. With the well-known Bayes theorem we give out a formula to calculate it and propose a method to discriminate a prior between prior distributions- Highest Posterior Belief (HPB). We also show that under certain condition, the HPB method is identical with the ML-II method.

关 键 词：贝叶斯学习 贝叶斯判别分析 先验分布 概率 先验信念比 

分 类 号：O212.8[理学—概率论与数理统计]