Design of a Compact Neural Network Based on An Optical Broadcast Architecture

Abstract

A novel, compact optoelectronic hardware neural network architecture based on an optical broadcast scheme is proposed and demonstrated. The basic cell in the system is composed of electronic neurons that share the same time-distributed, optical broadcast input. The system combines the computational strengths of electronics and the communication strengths of optics, and employs a modular reconfigurable architecture that is potentially scalable to a very large number of neurons while maintaining compactness. These characteristics are realized in an architecture that combines the integration of the processing elements in complementary metal-oxide semiconductor (CMOS) technology with the construction of efficient optical interconnection elements with focusing properties based on multiplexed volume holograms