Component Reduction for Gaussian Mixture Models

Kumiko MAEBASHI  Nobuo SUEMATSU  Akira HAYASHI  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E91-D   No.12   pp.2846-2853
Publication Date: 2008/12/01
Online ISSN: 1745-1361
DOI: 10.1093/ietisy/e91-d.12.2846
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Pattern Recognition
Keyword: 
mixture model,  EM-algorithm,  maximum likelihood,  hierarchical clustering,  

Full Text: PDF>>
Buy this Article




Summary: 
The mixture modeling framework is widely used in many applications. In this paper, we propose a component reduction technique, that collapses a Gaussian mixture model into a Gaussian mixture with fewer components. The EM (Expectation-Maximization) algorithm is usually used to fit a mixture model to data. Our algorithm is derived by extending mixture model learning using the EM-algorithm. In this extension, a difficulty arises from the fact that some crucial quantities cannot be evaluated analytically. We overcome this difficulty by introducing an effective approximation. The effectiveness of our algorithm is demonstrated by applying it to a simple synthetic component reduction task and a phoneme clustering problem.