Presentation 1997/3/17
The Leaning Ability of DBM with Quantized Synapses
Souichi Shibata, Shigeo Sato, Koji Nakajima,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we report a study on learning ability of DBM with quantized synapses in view of using analog memories for neurochips. We focused on XOR problem with a 2-2-1 DBM. Numerical simulations was done with a proper annealing schedule which was same as in case of using continuos synapses. Results show that the learning ability depends on how its synaptic weight is quantized and 10~11bit resolution of a quantized synaptic weight is required for successful learning.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) analog memory / neurochip / DBM / quantized synapse
Paper # NC96-131
Date of Issue

Conference Information
Committee NC
Conference Date 1997/3/17(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) The Leaning Ability of DBM with Quantized Synapses
Sub Title (in English)
Keyword(1) analog memory
Keyword(2) neurochip
Keyword(3) DBM
Keyword(4) quantized synapse
1st Author's Name Souichi Shibata
1st Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.()
2nd Author's Name Shigeo Sato
2nd Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.
3rd Author's Name Koji Nakajima
3rd Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.
Date 1997/3/17
Paper # NC96-131
Volume (vol) vol.96
Number (no) 583
Page pp.pp.-
#Pages 6
Date of Issue