Presentation	1999/6/18 Nonlinear Principal Component Analysis by MLP using Superposed Energy : On estimation of the intrinsic dimensionality of data Takashi TAKAHASHI, Ryuji TOKUNAGA,
PDF Download Page	PDF download Page Link
Abstract(in Japanese)	(See Japanese page)
Abstract(in English)	In order to obtain good performance in the autoassociative learning of a sandglass-type MLP, the dimensionality of its bottleneck-layer must be selected appropriately. This paper reports that a simple energy function, called "superposed energy," is applicable to nonlinear principal component analysis, and that the intrinsic dimensionality of nonlinear data is successfully estimated by applying the statistical criteria proposed by Hiraoka and Yoshizawa.
Keyword(in Japanese)	(See Japanese page)
Keyword(in English)	sandglass-type MLP / autoassociative learning / principal component analysis / dimensionality reduction
Paper #	NC99-27
Date of Issue

Conference Information
Committee	NC
Conference Date	1999/6/18(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)	Nonlinear Principal Component Analysis by MLP using Superposed Energy : On estimation of the intrinsic dimensionality of data
Sub Title (in English)
Keyword(1)	sandglass-type MLP
Keyword(2)	autoassociative learning
Keyword(3)	principal component analysis
Keyword(4)	dimensionality reduction
1st Author's Name	Takashi TAKAHASHI
1st Author's Affiliation	JSPS Research Fellow, Institute of Information Sciences and Electronics, University of Tsukuba()
2nd Author's Name	Ryuji TOKUNAGA
2nd Author's Affiliation	Institute of Information Sciences and Electronics, University of Tsukuba
Date	1999/6/18
Paper #	NC99-27
Volume (vol)	vol.99
Number (no)	131
Page	pp.pp.-
#Pages	8
Date of Issue