Title :
Bayesian approaches to Gaussian mixture modeling
Author :
Roberts, Stephen J. ; Husmeier, Dirk ; Rezek, Iead ; Penny, William
Author_Institution :
Dept. of Electr. & Electron. Eng., Imperial Coll. of Sci., Technol. & Med., London, UK
fDate :
11/1/1998 12:00:00 AM
Abstract :
A Bayesian-based methodology is presented which automatically penalizes overcomplex models being fitted to unknown data. We show that, with a Gaussian mixture model, the approach is able to select an “optimal” number of components in the model and so partition data sets. The performance of the Bayesian method is compared to other methods of optimal model selection and found to give good results. The methods are tested on synthetic and real data sets
Keywords :
Bayes methods; Hessian matrices; covariance matrices; parameter estimation; pattern clustering; unsupervised learning; Bayesian approaches; Gaussian mixture modeling; optimal model selection; Bayesian methods; Distributed computing; Equations; Parameter estimation; Probability density function; Roentgenium; Taylor series; Testing; Unsupervised learning;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
