Presentation 2005/3/23
On the Stochastic Complexity of Hidden Markov Models in Variational Bayesian Learning
Tikara HOSINO, Kazuho WATANABE, Sumio WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Variational Bayesian Learning is proposed for the approximation method of Bayesian learning. In spite of efficiency and experimental good performance, their mathematical property has not yet been clarified. In this paper we analyze variational Bayesian hidden Markov models which includes the true one thus the model is non-identifiable. We derive their asymptotic stochastic complexity. It is shown that in some prior condition, the stochastic complexity is much smaller than identifiable models.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Hidden Markov Models / Non-identifiability / Variational Bayesian Learning / Stochastic Complexity
Paper # NC2004-225
Date of Issue

Conference Information
Committee NC
Conference Date 2005/3/23(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 Neurocomputing (NC)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) On the Stochastic Complexity of Hidden Markov Models in Variational Bayesian Learning
Sub Title (in English)
Keyword(1) Hidden Markov Models
Keyword(2) Non-identifiability
Keyword(3) Variational Bayesian Learning
Keyword(4) Stochastic Complexity
1st Author's Name Tikara HOSINO
1st Author's Affiliation Computational Intelligence and System Science, Tokyo Inistitute of Technology:Nihon Unisys, Ltd.()
2nd Author's Name Kazuho WATANABE
2nd Author's Affiliation Computational Intelligence and System Science, Tokyo Inistitute of Technology
3rd Author's Name Sumio WATANABE
3rd Author's Affiliation Precision and Intelligence Laboratory, Tokyo Institute of Technology
Date 2005/3/23
Paper # NC2004-225
Volume (vol) vol.104
Number (no) 760
Page pp.pp.-
#Pages 6
Date of Issue