Presentation 1999/7/19
Model Comparisons and Predictions for Hierarchical Bayesian Neural Nets : Quadratic Approximation vs. MCMC
Y. Nakajima, M. Asano, Y. Nakada, T. Matsumoto,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) MCMC is used to perform model comparisons and compute predictive mean for Hierarchical Bayesian Neural Nets. The scheme is tested against two examples which show the fact that Quadratic Approximation for marginal likelihood is reasonably sound.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Quadratic Approximation / Hybrid Monte Carlo / Annealed Importance Sampling / Marginal likelihood / Model comparison
Paper # NC99-35
Date of Issue

Conference Information
Committee NC
Conference Date 1999/7/19(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) Model Comparisons and Predictions for Hierarchical Bayesian Neural Nets : Quadratic Approximation vs. MCMC
Sub Title (in English)
Keyword(1) Quadratic Approximation
Keyword(2) Hybrid Monte Carlo
Keyword(3) Annealed Importance Sampling
Keyword(4) Marginal likelihood
Keyword(5) Model comparison
1st Author's Name Y. Nakajima
1st Author's Affiliation Department of Electrical, Electronics and Computer Engineering, Waseda University()
2nd Author's Name M. Asano
2nd Author's Affiliation Department of Electrical, Electronics and Computer Engineering, Waseda University
3rd Author's Name Y. Nakada
3rd Author's Affiliation Department of Electrical, Electronics and Computer Engineering, Waseda University
4th Author's Name T. Matsumoto
4th Author's Affiliation Department of Electrical, Electronics and Computer Engineering, Waseda University
Date 1999/7/19
Paper # NC99-35
Volume (vol) vol.99
Number (no) 193
Page pp.pp.-
#Pages 7
Date of Issue