Presentation 2003/12/1
Piecewise Multivariate Polynomial Regression using Multilayer Perceptrons
Yusuke TANAHASHI, Takahiro ITO, Kazumi SAITO, Ryohei NAKANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Polynomials are quite suitable for expressing underlying regularities among multiple variates. When data have both numeric and nominal variates, we should solve piecewise multivariate polynomial regression. The regression means that we should automatically and adaptively divide the space spanned by nominal variates into subspaces,and within each subspace we should find the optimal multivarite regression function by using numeric variates. There has been no practical methods to solve such a hard problem. We have employed an numeric approach to the problem and proposed RF6.3 which is based on the finding that the problem can be decomposed into two subproblems: learning of a multilayer perceptron and rule restoration from the perceptron. We outline the method of RF6.3 and show how it can be applied to an estimation problem of professional baseball players salaries.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) multi-layer perceptron / piecewise polynomial regression / MCV regularizer / double layer of cross validation / rule restoration
Paper # NC2003-94
Date of Issue

Conference Information
Committee NC
Conference Date 2003/12/1(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) Piecewise Multivariate Polynomial Regression using Multilayer Perceptrons
Sub Title (in English)
Keyword(1) multi-layer perceptron
Keyword(2) piecewise polynomial regression
Keyword(3) MCV regularizer
Keyword(4) double layer of cross validation
Keyword(5) rule restoration
1st Author's Name Yusuke TANAHASHI
1st Author's Affiliation Nagoya Institute of Technology()
2nd Author's Name Takahiro ITO
2nd Author's Affiliation Nagoya Institute of Technology
3rd Author's Name Kazumi SAITO
3rd Author's Affiliation NTT Communication Science Laboratories, NTT Corporation
4th Author's Name Ryohei NAKANO
4th Author's Affiliation Nagoya Institute of Technology
Date 2003/12/1
Paper # NC2003-94
Volume (vol) vol.103
Number (no) 490
Page pp.pp.-
#Pages 5
Date of Issue