Presentation 2002/10/10
Algebraic Geometry of Complete Bipartite Graph-type Boltzmann Machines
Keisuke YAMAZAKI, Sumio WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) It is well known that Boltzmann machines are non-regular statistical models. The set of their parameters of small size models is an analystic set with singularities in the space of a large size ones. The mathematical foundation of their learning is not yet constructed because of the singularities, though they are applied in many situations of information engineering. Recently we established the method to calculate the bayes generalization errors with an algebraic geometric method even if the models are non-regular. This paper shows that the upper bounds of generalization errors in Boltzmann machines is smaller than the ones in regular statistical models.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Boltzmann machine / Generalization error / Algebraic Geometry
Paper # NC2002-53
Date of Issue

Conference Information
Committee NC
Conference Date 2002/10/10(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) Algebraic Geometry of Complete Bipartite Graph-type Boltzmann Machines
Sub Title (in English)
Keyword(1) Boltzmann machine
Keyword(2) Generalization error
Keyword(3) Algebraic Geometry
1st Author's Name Keisuke YAMAZAKI
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 Precidion and Intelligence Laboratory, Tokyo Institute of Technology
Date 2002/10/10
Paper # NC2002-53
Volume (vol) vol.102
Number (no) 381
Page pp.pp.-
#Pages 6
Date of Issue