摘要
贝叶斯线性判别是判别分析中流行的一种判别法,该方法在实现过程需要涉及高维样本协方差矩阵逆的复杂运算。本文利用近似贝叶斯方法对贝叶斯线性判别进行了算法设计和研究。该方法简单明了,易于实现并且规避了高维样本协方差矩阵逆的计算。本文还结合R软件对常见的分类案例给出近似贝叶斯计算和估计,从而说明了近似贝叶斯计算的简单有效性。
Bayesian linear discrimination is a popular method in discriminant analysis, which involves complex operation of the inverse of high dimensional sample covariance matrix in the realization process.In this paper, an approximate bayesian method is used to design and study bayesian linear discriminant.The method is simple and easy to implement and avoids the inverse calculation of high-dimensional sample covariance matrix.In this paper, the approximate bayesian computation and estimation of common classification cases with R software are presented, which indicates the simple validity of approximate bayesian computation.
出处
《电脑知识与技术》
2018年第10Z期203-204,207,共3页
Computer Knowledge and Technology
关键词
贝叶斯判别分析
近似贝叶斯计算
R软件
Bayes discriminant analysis
Approximate Bayesian computation
R software
