Presentation 2012-11-07
Model Selection of Bernoulli Mixture in Variational Bayes Learning
Kazumi DAKUJAKU, Sumio WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Variational Bayes learning, which is defined by the mean field approximation of the joint posterior distribution of a parameter and a hidden variable, approximates the Bayes posterior distribution with a low computational cost. In Bayes learning, it was proved that WAIC is an asymptotic unbiased estimator of the generalization loss, whereas in variational Bayes learning such an information criterion is still unknown. Recently a new method WAIC-VB was proposed which approximates the Bayes generalization loss using the importance sampling method based on the variational Bayes. In the present paper, we propose that WAIC-VB is useful in the model selection problem of a Bernoulli mixture, which is shown by experimental results in both artificial and practical applications.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) variational Bayes method / cross validation / Gaussian mixture / hyperparameter
Paper # IBISML2012-39
Date of Issue

Conference Information
Committee IBISML
Conference Date 2012/10/31(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Information-Based Induction Sciences and Machine Learning (IBISML)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Model Selection of Bernoulli Mixture in Variational Bayes Learning
Sub Title (in English)
Keyword(1) variational Bayes method
Keyword(2) cross validation
Keyword(3) Gaussian mixture
Keyword(4) hyperparameter
1st Author's Name Kazumi DAKUJAKU
1st Author's Affiliation Department of Computational Intelligence and Systems Science, Tokyo Institute of Technology()
2nd Author's Name Sumio WATANABE
2nd Author's Affiliation Department of Computational Intelligence and Systems Science, Tokyo Institute of Technology
Date 2012-11-07
Paper # IBISML2012-39
Volume (vol) vol.112
Number (no) 279
Page pp.pp.-
#Pages 6
Date of Issue