A non asymptotic penalized criterion for Gaussian mixture model selection - Archive ouverte HAL Access content directly
Reports (Research Report) Year : 2008

A non asymptotic penalized criterion for Gaussian mixture model selection

(1) , (1)
1
Cathy Maugis
Bertrand Michel

Abstract

Specific Gaussian mixtures are considered to solve simultaneously variable selection and clustering problems. A non asymptotic penalized criterion is proposed to choose the number of mixture components and the relevant variable subset. Because of the non linearity of the associated Kullback-Leibler contrast on Gaussian mixtures, a general model selection theorem for MLE proposed by Massart (2007) is used to obtain the penalty function form. This theorem requires to control the bracketing entropy of Gaussian mixture families. The ordered and non-ordered variable selection cases are both addressed in this paper.
Fichier principal
Vignette du fichier
RR-6549.pdf (393.76 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

inria-00284613 , version 1 (03-06-2008)
inria-00284613 , version 2 (04-06-2008)

Identifiers

  • HAL Id : inria-00284613 , version 2

Cite

Cathy Maugis, Bertrand Michel. A non asymptotic penalized criterion for Gaussian mixture model selection. [Research Report] RR-6549, INRIA. 2008. ⟨inria-00284613v2⟩
146 View
406 Download

Share

Gmail Facebook Twitter LinkedIn More