Accession Number : ADP008623
Title : Optical Modular Architectures for Multi-Layer BAM with 2-Dimensional Patterns,
Corporate Author : KOREA ADVANCED INST OF SCIENCE AND TECHNOLOGY SEOUL
Personal Author(s) : Lee, Soo-Young ; Lee, Hyuek-Jae ; Shin, Sang-Yung
Report Date : 22 MAY 1992
Pagination or Media Count : 4
Abstract : After the first demonstration of optically-implemented Hopfield model many neural network models have been investigated for large-scale optical implementation. The 1-dimensional Hopfield model has been extended for 2-dimensional patterns, and optical implementation of bidirectional associative memory (BAM) and quadratic associative memory had been investigated. Adaptive neural network models such as multi-layer perceptron has also been demonstrated. However performance of the simple Hopfield model and BAM is very limited, and many adaptive learning algorithms are too complicated to be implemented efficiently by optics. Also, when a new pattern need be added to the existing system, the correlation matrix learning rule of both the Hopfield model and BAM requires simple addition to existing interconnection weights, while error back-propagation learning rule for multi-layer perceptron requires to bring over all the previously stored patterns. Recently-we had extended the BAM into multi-layer architecture, of which performance is quite comparable to that of multi-layer perceptron. This multi-layer BAM (MBAM) still utilizes correlation matrices for easy optical implementation with outer-product matrix formation of inner-product recall. In this paper optical system architectures for the MBAM are presented for 2-dimensional patterns, and several implementation issues are discussed.
Descriptors : *OPTICAL STORAGE, *NEURAL NETS, *PATTERN RECOGNITION, PHOTODIODES, SEMICONDUCTOR LASERS, SEMICONDUCTOR DIODES, OPTICAL LENSES.
Subject Categories : Electrooptical and Optoelectronic Devices
Distribution Statement : APPROVED FOR PUBLIC RELEASE