上海交通大学学报(自然版) ›› 2013, Vol. 47 ›› Issue (07): 1119-1125.

• 无线电电子学、电信技术 • 上一篇    下一篇

混合高斯分布的变分贝叶斯学习参数估计

徐定杰,沈忱,沈锋   

  1. (哈尔滨工程大学 自动化学院,哈尔滨 150001)
     
  • 收稿日期:2012-07-13 出版日期:2013-07-30 发布日期:2013-07-30
  • 基金资助:

    国家自然科学基金资助项目(61102107),中国博士后基金项目(20100480979),黑龙江省博士后基金项目(LBHZ10193)

Variational Bayesian Learning for Parameter Estimation of Mixture of Gaussians

XU Dingjie,SHEN Chen,SHEN Feng
  

  1. (College of Automation, Harbin Engineering University, Harbin 150001, China)
  • Received:2012-07-13 Online:2013-07-30 Published:2013-07-30

摘要:

针对常用于非高斯信号或系统建模的包含隐变量的混合高斯分布模型,提出利用一种变分贝叶斯学习算法进行模型的参数估计.该方法采用一个形式较为简单的自由分布,通过不断最大化边缘似然函数的下界,迭代地更新变分参数,直至近似分布足够逼近参数真实的后验分布,从而实现混合高斯分布的参数估计.文中推导了该方法对混合高斯模型参数学习过程.实验表明,变分贝叶斯学习可以有效实现高斯混合模型的多参数估计,相比采样方法更有工程应用前景.
 
 

关键词: 参数估计, 混合模型, 高斯分布, 变分贝叶斯

Abstract:

Non-Gaussian signals or systems are usually modeled by mixture of Gaussians (MoG) models containing hidden variables. A variational Bayesian learning algorithm was suggested to infer the parameters of MoG. The algorithm estimateed the parameters of MoG by iteratively maximizing the lower bound of the marginal likelihood and updating the variational parameters until the free-form distribution was sufficiently close to the true posterior. The detailed learning of variational Bayes for MoG was derived and explained. The experiments show that this method can estimate the parameters of MoG favorably with sampling method from the engineering view.
 

Key words: parameters estimation, mixture model, Gaussian distribution, variational Bayes

中图分类号: