Presentation 1994/4/22
Weight Update Method for Neural Network LSI with Low Bit-Storage Synapses
Tomohisa Kimura, Takeshi Shima,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A synapse weight bit accuracy is reduced for constructing a large scale neural network on a limited small chip size.This paper describes a novel learning method for a neural network LSI which has low bit-storage synapse weights.In this method,the number of synapses which are modified at each learning iteration is limited. Simulation results show a learning per-formance is improved by taking the method.This proposed method is applicable to an LSI which has low bit-storage synapse weights.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Neural Network / LSI / Learning / Bit accuracy
Paper # CPSY94-21,FTS94-21,ICD94-21
Date of Issue

Conference Information
Committee CPSY
Conference Date 1994/4/22(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 Computer Systems (CPSY)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Weight Update Method for Neural Network LSI with Low Bit-Storage Synapses
Sub Title (in English)
Keyword(1) Neural Network
Keyword(2) LSI
Keyword(3) Learning
Keyword(4) Bit accuracy
1st Author's Name Tomohisa Kimura
1st Author's Affiliation Toshiba Research and Development Center()
2nd Author's Name Takeshi Shima
2nd Author's Affiliation Toshiba Research and Development Center
Date 1994/4/22
Paper # CPSY94-21,FTS94-21,ICD94-21
Volume (vol) vol.94
Number (no) 15
Page pp.pp.-
#Pages 8
Date of Issue