The Fundamental Problem of Gibbs Sampler in Mixture Models

Authors

  • Gholamhossein Gholami Department of Mathematics, Faculty of science, Urmia University, Urmia, IRAN
  • Hamideh Rasi Department of Mathematics, Urmia Branch, Islamic Azad University, Urmia, IRAN
  • Ali Etemadi Department of Statistics, Faculty of Mathematics, Tabriz University, Tabriz, IRAN

Keywords:

Gibbs sampler Mixture models Latent variable Posterior distribution Prior distribution

Abstract

The mixture models were rstly studied by Pearson in 1894. These models are strong tools,through which the complicated systems can be analyzed in a wide range of disciplines such as As-tronomy, Economics, Mechanics, etc. although the structure of these models is apparently simple,it is very complicated to obtain maximum likelihood estimators and Bayesian ones in particular andit needs to be approximated in most cases. In this paper, we apply the Gibbs Sampling in order toapproximate the Bayesian Estimator in Mixture models, present the Gibbs algorithms for the familyof exponential distributions and nally, we would show the disadvantage of this algorithm throughan example.

References

There are currently no refbacks

Published

2014-08-29

How to Cite

Gholami, G. ., Rasi, H. ., & Etemadi, A. . (2014). The Fundamental Problem of Gibbs Sampler in Mixture Models. Scientific Journal of Pure and Applied Sciences, 3(8), 809-815. Retrieved from http://sjournals.com/index.php/sjpas/article/view/590

Issue

Section

Mathematics

Most read articles by the same author(s)