Processing method, neural network structure and computer for simulating said neural network structure

摘要

The invention relates to a method of calculation of the standards used in a structure of the neural network or a computer. It makes it possible to determine the standard of the new synaptic coefficients without requiring the explicit calculation. For this purpose, the structure comprises a processing unit 10 which determines the output potentials yk and increments deltak, a storage unit 13 of the old and new standards, a calculation unit 15 of the new standards. Those - these are thus determined rapidly without requiring the use of the synaptic coefficients in the memory. The process can be used with some algorithms which require a calculation of the standards. / p & & p & application: neural networks.