Resumo

VIVAS, Hevert; MARTINEZ, Héctor Jairo  e  PEREZ, Rosana. Structured Secant Method for the Multilayer Perceptron Training. rev. cienc. [online]. 2014, vol.18, n.2, pp.131-150. ISSN 0121-1935.

In the group of models of artificial neural networks, it is the multilayer perceptron, a unidirectional neural network consisting of three or more layers, its training is done by an algorithm called backpropagation. In this work, we introduced the structured secant method for the training of multilayer perceptron and we compare its numerical performance with other methods widely used with the same purpose. Some numerical experiments show a good performance of this algorithm

Palavras-chave : Multilayer perceptron; structured secant method; training neural network; nonlinear least squares.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )