Presentation | 2007-11-19 Constructing error-correcting output coding using latent variable models Nobuhiko YAMAGUCHI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | The object of this paper is to construct a multi-class classifier by Error-Correcting Output Coding (ECOC) approach. The algorithm of the ECOC approach is divided into three steps: 1) dividing K classes into two super groups, 2) constructing multiple binary classifiers which separate the two super groups, and 3) combining the multiple binary classifiers. The ECOC approach constructs a classifier by sequentially performing the three steps. However, since these three steps are mutually related to each other, it is desirable to simultaneously perform the three steps with considering each other's steps. We therefore propose ECOC-LVM (ECOC using Latent Variable Models) which simultaneously performs the three steps. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Error-correcting output coding / Neural networks / Pattern classification / Latent variable models |
Paper # | NC2007-70 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2007/11/11(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) | Constructing error-correcting output coding using latent variable models |
Sub Title (in English) | |
Keyword(1) | Error-correcting output coding |
Keyword(2) | Neural networks |
Keyword(3) | Pattern classification |
Keyword(4) | Latent variable models |
1st Author's Name | Nobuhiko YAMAGUCHI |
1st Author's Affiliation | Faculty of Science and Engineering, Saga University() |
Date | 2007-11-19 |
Paper # | NC2007-70 |
Volume (vol) | vol.107 |
Number (no) | 328 |
Page | pp.pp.- |
#Pages | 5 |
Date of Issue |