Abstract

The next two decades will see major advances into the incorporation of intelligence into computational systems. Since throughput rates will continue to dominate systems issues, optics most likely will have a major role to play, and we would be remiss not to identify these opportunities at this early stage while most efforts are still focussed on algorithm development. Optical computing presents a highly parallel environment which is particularly adept at the rapid generation of Fourier transforms, correlation functions, and convolution. A natural extension of these capabilities is the optical implementation of search and sort algorithms, which form the backbone of symbolic computing. Associative processing, which is the general class of neural analogue computing systems whose computation structure is based on models of human thought and reasoning processes, may be ripe for optical processing due to the relative ease with which optics can realize the extensive interconnect networks so vital to such processing.

© 1985 Optical Society of America