Accession Number : ADA203796

Title :   Implementing Recurrent Back-Propagation on the Connection Machine.

Descriptive Note : Final rept.,

Corporate Author : NAVAL RESEARCH LAB WASHINGTON DC

Personal Author(s) : Deprit, E M

PDF Url : ADA203796

Report Date : 02 Dec 1988

Pagination or Media Count : 111

Abstract : Pineda's Recurrent back-Propagation algorithm for neural networks has been implemented on the Connection Machine, a massively parallel processor. Two fundamentally different graph architectures underlying the nets were tested-one based on arcs, the other on nodes. Confirming the predominance of communication over computation, performance measurements underscore the necessity to make connections the basic unit of representation. Comparisons between these graphs algorithms lead to important conclusions concerning the parallel implementation of neural nets in both software and hardware. Keywords include: Neural networks; Recurrent back-propagation; and Connection machine. (RH)

Descriptors :   *ARCHITECTURE, *COMPUTATIONS, *GRAPHS, *NEURAL NETS, *PARALLEL PROCESSORS, ALGORITHMS, COMMUNICATION AND RADIO SYSTEMS, COMPUTER PROGRAMS

Subject Categories : Computer Hardware

Distribution Statement : APPROVED FOR PUBLIC RELEASE