<title>Adaptive optical neural network for pattern classification on complex backgrounds</title>

Abstract

Laboratory of optical. pa.tt.ern recognition and neur&. networkG, Re9€io.rch Centre "S. I. VtvUov state Opti.cal.InLitut€,St . P.Ler9burg, Rui.a. ABSTRACTThe adaptive neural network model for pattern classification on the complex background is proposed. The method of the detection and extraction of the unknown changing patterns, based on the optical correlation is proposed. The optical implementation of that NN model, formed by placing the optical correlator in a linear resonator to realize the vectors comparison arid imae preparation, is discussed.