Presentation 2009-01-19
Experimental Study of Bayesian Learning using Langevin Equation in Singular Learing Machines
Taruhi IWAGAKI, Sumio WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Langevin equation implies an algorithm that could make samples from the stationary distribution of a biased random walk equivalent to the posterior distribution of Bayesian learning. Sampling with Langevin equation uses gradient information of the target distribution, therefore it is expected to be efficient than Metropolis algorithm especially in wide parameter space of singular learning machines like neural networks. In this paper, we will discuss experimental results and generalization errors of both Langevin algorithm and Metropolis algorithm for neural networks with practical dimension.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Singular Learning Machines / Langevein Equation / Fokker-Planck Equation / Baysian Learning
Paper # NC2008-88
Date of Issue

Conference Information
Committee NC
Conference Date 2009/1/12(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) Experimental Study of Bayesian Learning using Langevin Equation in Singular Learing Machines
Sub Title (in English)
Keyword(1) Singular Learning Machines
Keyword(2) Langevein Equation
Keyword(3) Fokker-Planck Equation
Keyword(4) Baysian Learning
1st Author's Name Taruhi IWAGAKI
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 Precision and Intelligence Laboratory, Tokyo Institute of Technology
Date 2009-01-19
Paper # NC2008-88
Volume (vol) vol.108
Number (no) 383
Page pp.pp.-
#Pages 6
Date of Issue