Presentation 1995/6/16
Performance of Hierarchical Fractal Neural Net
B. Chakraborty, Y. Sawada,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Hierarchical nets having sparse and localized connectivity with fractal connections within layers have been studied. The performance of the proposed net in classification problem has been compared to that of fully connected multilayer perceptron and randomly connected sparse net with artificially generated fractal data set and a real data set derived from sonar signals for underwater target recognition. A simple version of backpropagation algorithm has been used to train all the nets. Fractal net seems to be far better than randomly connected sparse nets in fractal pattern recognition. For the second data set fractally connected nets performs well compared to fully connected net as fractal dimension is increased above 0. 75. Moreover fractal net seems to possess more generalization capability compared to fully connected net in recognizing patterns other than training patterns.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Hierarchical net / Fractal connection / Sparse net / Classification
Paper #
Date of Issue

Conference Information
Committee NLP
Conference Date 1995/6/16(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 Nonlinear Problems (NLP)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Performance of Hierarchical Fractal Neural Net
Sub Title (in English)
Keyword(1) Hierarchical net
Keyword(2) Fractal connection
Keyword(3) Sparse net
Keyword(4) Classification
1st Author's Name B. Chakraborty
1st Author's Affiliation Research Center for Electrical Communication Tohoku University()
2nd Author's Name Y. Sawada
2nd Author's Affiliation Research Center for Electrical Communication Tohoku University
Date 1995/6/16
Paper #
Volume (vol) vol.95
Number (no) 92
Page pp.pp.-
#Pages 7
Date of Issue