Stochastic model selection for Mixtures of Matrix-Normals

TL;DR

This paper introduces a Bayesian approach for simultaneous model estimation and selection in finite mixtures of matrix normal distributions, enhancing classification of three-way data beyond traditional EM methods.

Contribution

The paper develops a Bayesian framework for mixture model selection, offering a new tool for estimating the number of components in matrix normal mixture models.

Findings

Effective in simulation studies

Successfully applied to real data example

Improves model selection accuracy

Abstract

Finite mixtures of matrix normal distributions are a powerful tool for classifying three-way data in unsupervised problems. The distribution of each component is assumed to be a matrix variate normal density. The mixture model can be estimated through the EM algorithm under the assumption that the number of components is known and fixed. In this work we introduce, develop and explore a Bayesian analysis of the model in order to provide a tool for simultaneous model estimation and model selection. The effectiveness of the proposed method is illustrated on a simulation study and on a real example.