Geodesic
A non asymptotic penalized criterion for gaussian mixture model selection
ESAIM: Probability and Statistics, Tome 15 (2011), pp. 41-68

Voir la notice de l'article provenant de la source Numdam

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 maximum likelihood estimation proposed by [Massart Concentration inequalities and model selection Springer, Berlin (2007). Lectures from the 33rd Summer School on Probability Theory held in Saint-Flour, July 6-23 (2003)] 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.

DOI : 10.1051/ps/2009004
Classification : 62H30, 62G07
Keywords: model-based clustering, variable selection, penalized likelihood criterion, bracketing entropy 
@article{PS_2011__15__41_0,
     author = {Maugis, Cathy and Michel, Bertrand},
     title = {A non asymptotic penalized criterion for gaussian mixture model selection},
     journal = {ESAIM: Probability and Statistics},
     pages = {41--68},
     publisher = {EDP-Sciences},
     volume = {15},
     year = {2011},
     doi = {10.1051/ps/2009004},
     mrnumber = {2870505},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.1051/ps/2009004/}
}
TY  - JOUR
AU  - Maugis, Cathy
AU  - Michel, Bertrand
TI  - A non asymptotic penalized criterion for gaussian mixture model selection
JO  - ESAIM: Probability and Statistics
PY  - 2011
SP  - 41
EP  - 68
VL  - 15
PB  - EDP-Sciences
UR  - http://geodesic.mathdoc.fr/articles/10.1051/ps/2009004/
DO  - 10.1051/ps/2009004
LA  - en
ID  - PS_2011__15__41_0
ER  - 
%0 Journal Article
%A Maugis, Cathy
%A Michel, Bertrand
%T A non asymptotic penalized criterion for gaussian mixture model selection
%J ESAIM: Probability and Statistics
%D 2011
%P 41-68
%V 15
%I EDP-Sciences
%U http://geodesic.mathdoc.fr/articles/10.1051/ps/2009004/
%R 10.1051/ps/2009004
%G en
%F PS_2011__15__41_0
Maugis, Cathy; Michel, Bertrand. A non asymptotic penalized criterion for gaussian mixture model selection. ESAIM: Probability and Statistics, Tome 15 (2011), pp. 41-68. doi: 10.1051/ps/2009004

Cité par Sources :