SHIFT-INVARIANT INTERPATTERN ASSOCIATION NEURAL-NETWORK

Authors
UANG CM YIN SZ ANDRES P REESER W YU FTS
Citation
Cm. Uang et al., SHIFT-INVARIANT INTERPATTERN ASSOCIATION NEURAL-NETWORK, Applied optics, 33(11), 1994, pp. 2147-2151
Citations number
12
Categorie Soggetti
Optics
Journal title
Applied optics → ACNP
ISSN journal
00036935
Volume
33
Issue
11
Year of publication
1994
Pages
2147 - 2151
Database
ISI
SICI code
0003-6935(1994)33:11<2147:SIAN>2.0.ZU;2-K
Abstract
A shift-invariant neural network that uses the translation-invariant p roperty of the modulus Fourier spectra with the heteroassociation inte rpattern association memory is proposed. A binary encoding of a spectr al sampling of the training set is used to preserve the main features. Computer simulations and experimental demonstrations are provided tha t show the shift-invariant property of the proposed optical neural net work.