ON THE PROPERTIES OF THE FEEDFORWARD METHOD - A SIMPLE TRAINING LAW FOR ON-CHIP LEARNING

This paper investigates the properties of the so-called feedforward me thod, which is a very simple training law suitable for on-chip learnin g, Its merit is conceptual and implementational simplicity, Its signal s do not propagate in both directions and it works for various types o f activation function, a feature that makes it particularly effective in the case of unmodeled activation functions, Extensive simulation ha s shown that this method is usually faster than backpropagation.