Neural network device and method for data classification.

摘要

The device comprises a neural network with an input layer 3, an inner layer 4, an output layer 5. This network is intended for classifying data vectors according to classes, the synaptic weights in the network being determined by learning on the basis of specimens whose class is known. According to the invention, while learning, each class is defined as corresponding to a set of neurons each of which represent a domain which contain a fixed number of specimens. The network includes a number of neurons and synaptic weights which have been determined on the basis of the classes thus defined. Application, among others, to authentication of signatures. <IMAGE>