文档介绍：works,Vol. 10, No. 6, pp. 1143-1151, 1997
01997 Elsevier Science Ltd. All rightsreserved
Pergamon Printedin CheatBritain
@ 0893–6080/97 $+.03
PII: S0893-6080(96)00128-1
CONTRIBUTEDARTICLE
ically Trained Cellular works
MICHELE ZAMPARELLI
HLRZ-KFA 52425, and Dipartimento di Matematica, Universit6 degli Studi di Roma, “La Sapienza”
(Received6 June1995;accepted10 November1996)
Abstract—Real-ic algorithms on a parallel architecture are applied to optimize the synaptic couplings of a
Cellular work for specific greyscale image processing tasks. Using supervised learning information in the
jitnessfinction, we propose the ic Algorithm as a general training methodfor Cellular works. 01997
Elsevier Science Ltd.
Keywords-Cellular works,icalgorithms,Supervisedlearning,Imageprocessing.
1. INTRODUCTION system and its parallelization. ments on the
results are given in Section 7.
Cellular works (CNNS),a new type of locally
connected work with continuous activation
values, have recently demonstrated their efficacy for 2. THE CELLULAR WORK MODEL
bipolar signal processing. Several models of cortical neu-
Cellular works invented by Chua and Yang
rons have been proposed so far, but the time evolution in
(1988), consist of a partial unification of the paradigms
the neuron’s activation has always been rather under-
Cellular Automaton and work, retaining