Generalization performance of exchange monte carlo method for normal mixture models

Authors:
Kenji Nagata;Sumio Watanabe
Affiliations:
Department of Computational Intelligence and Systems Science, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan;P&I Lab.,Tokyo Institute of Technology
Venue:
IDEAL'06 Proceedings of the 7th international conference on Intelligent Data Engineering and Automated Learning
Year:
2006

Citing 2
Cited 0

Singularities in mixture models and upper bounds of stochastic complexity

Neural Networks
Algebraic Analysis for Nonidentifiable Learning Machines

Neural Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

A normal mixture model, which belongs to singular learning machines, is widely used in statistical pattern recognition. In singular learning machines, the Bayesian learning provides the better generalization performance than the maximum likelihood estimation. However, it needs huge computational cost to realize the Bayesian posterior distribution by the conventional Monte Carlo method. In this paper, we propose that the exchange Monte Carlo method is appropriate for the Bayesian learning in singular learning machines, and experimentally show that it provides better generalization performance in the Bayesian learning of a normal mixture model than the conventional Monte Carlo method.