Presentation | 2005/3/23 Localized Bayesian Learning for Singular Learning Machines Shingo TAKAMATSU, Sumio WATANABE, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In singular learning machines, like a three-layer neural network and a normal mixture, it is known that the learning method by mixture of many parameters, like Bayesian estimasion, is better than the method by deciding one position of parameters. But it is usually difficult to make the posterior distribution because these learning machines have their singularities in the parameter spaces. In this paper, we propose a new learning method with the localized posterior distribution and show that its generalization error can be equal to or better than the one of Bayesian estimation by applying this method to the reduced rank regression. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Reduced rank regression / Gerenalization error / Bayexian estimate |
Paper # | NC2004-227 |
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) | Localized Bayesian Learning for Singular Learning Machines |
Sub Title (in English) | |
Keyword(1) | Reduced rank regression |
Keyword(2) | Gerenalization error |
Keyword(3) | Bayexian estimate |
1st Author's Name | Shingo TAKAMATSU |
1st Author's Affiliation | Department of Computer Science Tokyo Institute of Technology() |
2nd Author's Name | Sumio WATANABE |
2nd Author's Affiliation | PI Lab., Tokyo Institute of Technology |
Date | 2005/3/23 |
Paper # | NC2004-227 |
Volume (vol) | vol.104 |
Number (no) | 760 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |