Share Email Print
cover

Proceedings Paper

Adaptive neural network for pattern recognition
Author(s): Eugene I. Shubnikov
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

A three-layered neural network for pattern recognition with feedback and complex states of neurons and interconnections is suggested. It consists of comparison, recognition, and selective attention layers. Comparison is realized in spectral space, recognition and selective attention are realized in image space. The recognition layer works as `winner takes all.' Parallel-sequential accessing to long term memory is used. Adaptation is realized by creation of new recognition categories and by long term memory change when input patterns are similar enough. A joint transform correlator with dynamic holographic filter is used in optical realization of adaptive resonance neural network.

Paper Details

Date Published: 21 January 1994
PDF: 9 pages
Proc. SPIE 2051, International Conference on Optical Information Processing, (21 January 1994); doi: 10.1117/12.166016
Show Author Affiliations
Eugene I. Shubnikov, S. I. Vavilov State Optical Institute (Russia)


Published in SPIE Proceedings Vol. 2051:
International Conference on Optical Information Processing
Yuri V. Gulyaev; Dennis R. Pape, Editor(s)

© SPIE. Terms of Use
Back to Top