SNAP: a parallel processor for implementing real-time neural networks

Author

Wojciechowski, Edward

Author_Institution

ITT Avionics, Nutley, NJ, USA

fYear

1991

fDate

20-24 May 1991

Firstpage

736

Abstract

The author describes the Scalable Neural Array Processor (SNAP), a single-instruction/multiple data (SIMD) parallel processing architecture for solving artificial neural systems, including Hopfield networks, adaptive resonance theory networks, multi-layer perceptron with back error propagation, and bidirectional associative memories. SNAP offers a low-cost, viable, neural network processing engine for real-time applications and research

Keywords

cellular arrays; content-addressable storage; neural nets; parallel architectures; real-time systems; Hopfield networks; SIMD; SNAP; Scalable Neural Array Processor; adaptive resonance theory networks; artificial neural systems; bidirectional associative memories; brassboard; cellular array; multi-layer perceptron with back error propagation; neural network processing engine; parallel processor; real-time neural networks; single layer nearest neighbor encoder; single-instruction/multiple data; Application software; Artificial neural networks; Biological neural networks; Computational modeling; Computer architecture; Computer networks; Computer simulation; Humans; Neural networks; Neurons;

fLanguage

English

Publisher

ieee

Conference_Titel

Aerospace and Electronics Conference, 1991. NAECON 1991., Proceedings of the IEEE 1991 National

Conference_Location

Dayton, OH

Print_ISBN

0-7803-0085-8

Type

conf

DOI

10.1109/NAECON.1991.165834

Filename

165834