Presentation 2003/3/10
Studies on An Redundant Units Discrimination Based on Geometrically Viewpoint
Naoki HIGA, Hiroshi SHIRATSUCHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This report studies on an additional pruning algorithm with structure learning with forgetting(SLF) for the redundant units of neural networks. Two types redundancy criteria of hidden units are considered : differences between two hidden unit outputs, and using separation hyperplane's angle, norm and distance. Through the analysis of similarity progresses in SLF, in case of large number of initial hidden units, it is shown that the redundant hidden units are still remain at the end of learning. Proposal learning algorithm eliminates redundant the hidden units, which have the high similarity between the hidden units, from the neural networks. From simulation results performed on iris classification problem. it was confirmed that the proposal learning algorithm give almost same network structure even if the number of the initial hidden units are increases.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Similarity / Pruning / Structural Learning with Forgetting / Separating Hyper plane / Structure Determination
Paper # NC2002-142
Date of Issue

Conference Information
Committee NC
Conference Date 2003/3/10(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) Studies on An Redundant Units Discrimination Based on Geometrically Viewpoint
Sub Title (in English)
Keyword(1) Similarity
Keyword(2) Pruning
Keyword(3) Structural Learning with Forgetting
Keyword(4) Separating Hyper plane
Keyword(5) Structure Determination
1st Author's Name Naoki HIGA
1st Author's Affiliation Faculty of Engineering, University of the Ryukyus()
2nd Author's Name Hiroshi SHIRATSUCHI
2nd Author's Affiliation Faculty of Engineering, University of the Ryukyus
Date 2003/3/10
Paper # NC2002-142
Volume (vol) vol.102
Number (no) 729
Page pp.pp.-
#Pages 6
Date of Issue