Presentation 2007-03-16
An Application to Pattern Generation Based on Backward Projection of Self-Organized Feature Map : In Case of Face Mark 'Emoticon'
Takahiro ISHIGUMA, Hiroshi WAKUYA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A major feature of the self-organizing map (SOM) is a topology-preserving projection between the input and competitive layers. Based on its distinctive nature, a new method for generating patterns through backward projection is proposed recently. In order to carry out further consideration, two kinds of revisions are tried. One is projecting backward from same firing area, not a point, in the competitive layer to the input layer, and the other is producing an animation as a series of backward-projected patterns along with a pathway between two training data. As a result of computer simulations, both of them show good performance and it is also confirmed their effectiveness.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) backward projection / pattern generation / self-organizing map / emoticon
Paper # NC2006-209
Date of Issue

Conference Information
Committee NC
Conference Date 2007/3/9(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) An Application to Pattern Generation Based on Backward Projection of Self-Organized Feature Map : In Case of Face Mark 'Emoticon'
Sub Title (in English)
Keyword(1) backward projection
Keyword(2) pattern generation
Keyword(3) self-organizing map
Keyword(4) emoticon
1st Author's Name Takahiro ISHIGUMA
1st Author's Affiliation Faculty of Science and Engineering, Saga University()
2nd Author's Name Hiroshi WAKUYA
2nd Author's Affiliation Faculty of Science and Engineering, Saga University
Date 2007-03-16
Paper # NC2006-209
Volume (vol) vol.106
Number (no) 590
Page pp.pp.-
#Pages 5
Date of Issue