Presentation 1994/10/13
Construction of Classification Trees Based on Features Provided by Random MLP
Qiangfu Zhao,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recently,the author has shown that random multilayer perceptrons with cross-layer connections (CLC-MLP) could be used as general purpose feature extractors,and complex patterns could be mapped into linearly separable ones.To gain some insight into the properties of random CLC-MLP,this paper investigates the goodness of features provided by random CLC-MLP using classification trees. A simple method is first introduced to construct binary trees for recognition of binary image patterns.Then,the classification tree approach is applied to invariant recognition of numerics (0-8). Experimental results show that useful features can be extracted automatically by using random CLC-MLP,and the size of classification trees can be greatly reduced.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Multilayer perceptron / Classification tree / Cross-Layer Connections / Feature extraction / Pattern recognition
Paper # NC94-35
Date of Issue

Conference Information
Committee NC
Conference Date 1994/10/13(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Neurocomputing (NC)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Construction of Classification Trees Based on Features Provided by Random MLP
Sub Title (in English)
Keyword(1) Multilayer perceptron
Keyword(2) Classification tree
Keyword(3) Cross-Layer Connections
Keyword(4) Feature extraction
Keyword(5) Pattern recognition
1st Author's Name Qiangfu Zhao
1st Author's Affiliation Graduate School of Information Sciences,Tohoku University()
Date 1994/10/13
Paper # NC94-35
Volume (vol) vol.94
Number (no) 272
Page pp.pp.-
#Pages 7
Date of Issue