<title>Hybrid optical neural computing</title>

Abstract

ABSTRACT Combination of Digital parallel computing and optical neural computing give us a new types of hybring computing. A temporal coding technique is employed for preprocessing digitalcomputing. This data is subjected by a hree-layer optical neural computing system to make character recognition. 1. INTRODUCTION Some efforts to develop optical computing systems have been made in this decade. Optical digital,optical analog and neural computing architectures have been discussed. These architectures have their own capabilities and possibilities in information processing. For example, the optical digital architecture is suitable for parallel logical and image processing[1,2}, whereas the optical neural computing for patternclassification and recognition[3,4]. Here an optical computing system combined with a digital system and aneural computing system is proposed. In the digital parallel processing system based on a temporal coding method, edge detection of binary input patterns and feature extraction is made, and then patternclassification is done by a three layer neural network system based on the back-propagation learning