Share Email Print

Proceedings Paper

Scheme and technical implementations of highly effective vector-matrix multipliers for an optoelectronic large-scale neurocoprocessor
Author(s): Makar V. Gunko; Oleg V. Rozhkov; Nina M. Verenikina
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Various scheme and technical implementations of a vector- matrix multiplier for neuronetworks design are considered. Variants of the VMM laboratory model on the basis of the traditional optics and standard optoelectronic components are worked out. Input and output vectors are realized accordingly as radiation source and photodetector matrices; weights set with the help of a projector LCD. Possibility of an integrated optical VMM implementation as a monolithic circuit is considered; peculiarity of a planar circuit working is noted. The results of circuit elements calculations and complete VMM circuit arranging with a control block diagram are presented. This neurochip is made on the GaAs substrate by means of the traditional semiconductor technology (0.35 mkm) and takes 40 X 10 mm2 area.

Paper Details

Date Published: 1 April 1998
PDF: 12 pages
Proc. SPIE 3402, Optical Information Science and Technology (OIST97): Optical Memory and Neural Networks, (1 April 1998); doi: 10.1117/12.304955
Show Author Affiliations
Makar V. Gunko, Moscow State Technical Univ. (Russia)
Oleg V. Rozhkov, Moscow State Technical Univ. (Russia)
Nina M. Verenikina, Moscow State Technical Univ. (Russia)

Published in SPIE Proceedings Vol. 3402:
Optical Information Science and Technology (OIST97): Optical Memory and Neural Networks
Andrei L. Mikaelian, Editor(s)

© SPIE. Terms of Use
Back to Top