FIELD: information technologies.
SUBSTANCE: method of intellectual information processing in a neuron network, which consists in supplying a signal into a double-layer network with feedbacks, and the signal is divided into components in the basis aligned with an input layer of the network, signal representation in the form of serial combinations of single images in compliance with the preset rules of its recognition with account of reverse recognition results, shifts of single images combinations along the layers with account of their current conditions, memorising results of recognition on network elements, using serial combinations of single images as processing results on the output layer of the network after the reverse conversion into corresponding initial signals, besides, shifts of combinations of single images along the layers are carried out with varied parameters of shifts, and combinations of single images are moved on along the layers following a spiral with variable diameter.
EFFECT: increased intelligence of information processing, improved memorising, recognition and extraction of signals from the network.
7 dwg, 1 tbl
