Título: Aceleração Da Convergência Do Algoritmo De Treinamento SMC-BP Através Do Ajuste Da Superfície De Deslizamento
Autores: Parma, Gustavo G.; Braga, Antônio P.; Menezes, Benjamim R.
Resumo: An improved algorithm based on the SMC-BP [3] algorithm using the well known sliding mode theory is proposed. The sliding mode surface is obtained from the squared error of the training set and then the weight update equations were derived using the gradient’s addition. The bounds for the gains (learning rates) are obtained from the sliding surface so that system stability is guaranteed. The proposed algorithm is applied to function approximation problems and results are compared with the standard backpropagation and Rprop algorithms in terms of speed of learning.
Palavras-chave:
Páginas: 4
Artigo em pdf: 4cbrn_040.pdf
Arquivo BibTex: 4cbrn_040.bib