Accession Number : ADP008621

Title :   Optical Matrix - Vector Implementation of Binary Valued Backpropagation,

Corporate Author : CALIFORNIA UNIV SAN DIEGO LA JOLLA DEPT OF ELECTRICAL AND COMPUTER ENGINEERIN G

Personal Author(s) : Brodsky, Stephen A. ; Guest, Clark C.

Report Date : 22 MAY 1992

Pagination or Media Count : 4

Abstract : Optical implementation of neural networks can combine advantages of neural network adaptive parallel processing and optical free-space connectivity. binary valued Backpropagation, a supervised learning algorithm related to standard Backpropagation, significantly reduces interconnection storage and computation requirements. This implementation of binary valued Backpropagation used optical matrix-vector multiplication to represent the forward information flow between network layers. Previous analog optical network memory systems have been described.

Descriptors :   *OPTICAL PROCESSING, *NEURAL NETS, OPTICAL STORAGE, PARALLEL PROCESSING, BINARY ARITHMETIC, ALGORITHMS, LIGHT MODULATORS, BEAM SPLITTING.

Subject Categories : Electrooptical and Optoelectronic Devices
      Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE