VLSI design of optically coupled neural networks

Author

Chandna, A. ; Jullien, G.A. ; Miller, W.C.

Author_Institution

Windsor Univ., Ont., Canada

fYear

1989

fDate

14-16 Aug 1989

Firstpage

493

Abstract

An initial study into the construction of photo receptors and photo-activated synapses for use in massively parallel analog CMOS neural networks is discussed. With this work the authors plan to construct circuits that do not rely on limited input pins for operation of a neural network, but rather use light as both the input and programming mechanism. Experiments are conducted with two simple circuits: a photo receptor that logarithmically relates output voltage to input light intensity and a floating optically programmed resistor that is used as a synaptic weight in a resistor network summer input to a neuron amplifier

Keywords

CMOS integrated circuits; VLSI; analogue computer circuits; integrated optoelectronics; neural nets; parallel processing; VLSI design; analog CMOS neural networks; floating optically programmed resistor; massively parallel neural networks; neuron amplifier; optically coupled neural networks; photo-activated synapses; photoreceptors; programming mechanism; resistor network summer input; synaptic weight; Coupling circuits; Neural networks; Optical computing; Optical coupling; Optical design; Optical fiber networks; Pins; Resistors; Very large scale integration; Voltage;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1989., Proceedings of the 32nd Midwest Symposium on

Conference_Location

Champaign, IL

Type

conf

DOI

10.1109/MWSCAS.1989.101899

Filename

101899