Presentation 2005/3/23
The Accuracy of Mean Field Approximation as an Approach of the True Bayesian Learning
Nobuhiro NAKANO, Sumio WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Mean field approximation is proposed as an approach of the Bayesian learning in order to overcome intractability of computing the Bayesian posterior distributions. However, it remains unknown how precise mean field approximation in singular learning machines is. In this paper, we consider a case when a three-layer perceptron that includes the true distributionis is trained to estimate the true one, and derive its asymptotic stochastic complexity that enables us to discuss the accuracy of mean field approximation as an approach of the true Bayesian learning.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Algebraic Geometry / Stochastic Complexity / Learning Machines with Singularities / Mean Field Approximation
Paper # NC2004-212
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 JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) The Accuracy of Mean Field Approximation as an Approach of the True Bayesian Learning
Sub Title (in English)
Keyword(1) Algebraic Geometry
Keyword(2) Stochastic Complexity
Keyword(3) Learning Machines with Singularities
Keyword(4) Mean Field Approximation
1st Author's Name Nobuhiro NAKANO
1st Author's Affiliation Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology()
2nd Author's Name Sumio WATANABE
2nd Author's Affiliation P&I Lab., Tokyo Institute of Technology
Date 2005/3/23
Paper # NC2004-212
Volume (vol) vol.104
Number (no) 760
Page pp.pp.-
#Pages 6
Date of Issue