正态判别中完全贝叶斯和贝叶斯最小二乘准则的比较

Comparison of Full Bayes and Bayes-Least Square Criteria for Normal Discrimination

  • 摘要: 本文研究了两个具相同协方差阵的多元正态总体在自然共轭九验分布下的判别分析问题,证明了当可观察变量的个数趋无穷时,贝叶斯判别函数与一基于样本的判别函数渐近等价。

     

    Abstract: This paper investigates the problem of discrimination between two normal populations with common dispersion matrix, using a natural conjugate prior. It is shown that the Bayesian discriminant is asymptotically equivalent to a sample-based discriminant when the number of observed variables is allowed to tend to infinity.

     

/

返回文章
返回