Title :
A successive overrelaxation backpropagation algorithm for neural-network training
Author :
De Leone, Renato ; Capparuccia, Rosario ; Merelli, Emanuela
Author_Institution :
Dipartimento di Matematica e Fisica, Camerino Univ., Macerata, Italy
fDate :
5/1/1998 12:00:00 AM
Abstract :
A variation of the classical backpropagation algorithm for neural network training is proposed, and convergence is established using the perturbation results of Mangasarian and Solodov (1994). The algorithm is similar to the successive overrelaxation (SOR) algorithm for systems of linear equations and linear complementary problems in using the most recently computed values of the weights to update the values on the remaining arcs
Keywords :
backpropagation; convergence of numerical methods; neural nets; optimisation; perturbation techniques; backpropagation; convergence; learning; linear complementary problems; neural-network; optimisation; perturbation; successive overrelaxation; Backpropagation algorithms; Biological neural networks; Boolean functions; Convergence; Equations; Humans; Network topology; Pattern recognition; Proteins; Speech recognition;
Journal_Title :
Neural Networks, IEEE Transactions on
