A large scale neural network with nonlinear synapses
J.B. Lont · 1994
In this paper, we prove the usefulness of multi-layer perceptrons with nonlinear synapses for real-size applications such as handwritten character recognition. We describe some simulations of a large scale neural network with nonlinear synapses. The convergence speed is shown to be better than for the networks with linear multipliers. Some issues concerning the VLSI implementation are discussed.>