Penalized Maximum Likelihood Estimator for Normal Mixtures
Authors:
- 28 February 2003
- journal article
- Published by Wiley in Scandinavian Journal of Statistics
- Vol. 30 (1) , 45-59
Abstract
The estimation of the parameters of a mixture of Gaussian densities is considered, within the framework of maximum likelihood. Due to unboundedness of the likelihood function, the maximum likelihood estimator fails to exist. We adopt a solution to likelihood function degeneracy which consists in penalizing the likelihood function. The resulting penalized likelihood function is then bounded over the parameter space and the existence of the penalized maximum likelihood estimator is granted. As original contribution we provide asymptotic properties, and in particular a consistency proof, for the penalized maximum likelihood estimator. Numerical examples are provided in the finite data case, showing the performances of the penalized estimator compared to the standard one.Keywords
This publication has 25 references indexed in Scilit:
- Bayesian analysis of mixture models with an unknown number of components—an alternative to reversible jump methodsThe Annals of Statistics, 2000
- Bayesian approaches to Gaussian mixture modelingIEEE Transactions on Pattern Analysis and Machine Intelligence, 1998
- Bayesian Density Estimation and Inference Using MixturesJournal of the American Statistical Association, 1995
- A constrained em algorithm for univariate normal mixturesJournal of Statistical Computation and Simulation, 1986
- A Constrained Formulation of Maximum-Likelihood Estimation for Normal Mixture DistributionsThe Annals of Statistics, 1985
- Some Methods for Studying the Mixture of Two Normal (Lognormal) DistributionsJournal of the American Statistical Association, 1979
- Nonparametric Roughness Penalties for Probability DensitiesBiometrika, 1971
- A NOTE ON THE CONSISTENCY AND MAXIMA OF THE ROOTS OF LIKELIHOOD EQUATIONSBiometrika, 1954
- On Wald's Proof of the Consistency of the Maximum Likelihood EstimateThe Annals of Mathematical Statistics, 1949
- A Generalized Theory of the Combination of Observations so as to Obtain the Best ResultAmerican Journal of Mathematics, 1886