Proceedings of the Fuzzy System Symposium
24th Fuzzy System Symposium
Session ID : TF4-3
Conference information

Host: Japan Society for Fuzzy Theory and Intelligent Informatics

A Trial of On-line Character Recognition with a Feedback SOM --- Part-II Revision of Training Method for Large Data Set ---
*Hiroshi Wakuya, Toshiro Noda
Author information
Keywords: self-organizing map, feedback SOM, time series, character recognition, training method
CONFERENCE PROCEEDINGS FREE ACCESS

Details
Download PDF (1231K)
Download citation RIS

(compatible with EndNote, Reference Manager, ProCite, RefWorks)

BIB TEX

(compatible with BibDesk, LaTeX)

Text
How to download citation
Contact us
Abstract
A feedback SOM is a revised version of the standard SOM for dealing with time-dependent data. Since its proposal, even though its effectiveness has been confirmed, most applications are carried out with small data sets. Then, in this article, an on-line character recognition task with about 20 samples is tried. As a result, it is found experimentally that revision of training method for time series is effective to avoid local reinforcement of the initial state and its neighborhood, which led the SOM to similar responses in spite of different input patterns.
Content from these authors
© 2008 Japan Society for Fuzzy Theory and Intelligent Informatics
Previous article Next article
Favorites & Alerts

Recently viewed articles
Share this page
feedback
 Top