Stochastic Approximation and Multilayer Perceptrons: The Gain Backpropagation Algorithm
P.J. Gawthrop
D. Sbarbaro
Department of Mechanical Engineering, The University,
Glasgow G12 8QQ, United Kingdom
Abstract
A standard general algorithm, the stochastic approximation algorithm of Albert and Gardner [1], is applied in a new context to compute the weights of a multilayer perceptron network. This leads to a new algorithm, the gain backpropagation algorithm, which is related to, but significantly different from, the standard backpropagation algorithm [2]. Some simulation examples show the potential and limitations of the proposed approach and provide comparisons with the conventional backpropagation algorithm.
