An analytical framework for local feedforward networks

Interference in neural networks occurs when learning in one area of the input space causes unlearning in another area. Networks that are less susceptible to interference are referred to as spatially local networks. To obtain a better understanding of these properties, a theoretical framework, consisting of a measure of interference and a measure of network localization, is developed. These measures incorporate not only the network weights and architecture but also the learning algorithm. Using this framework to analyze sigmoidal,multi-layerperceptron (MLP) networks that employ the back-propagation learning algorithm on the quadratic cost function, we address a familiar misconception that single-hidden-layer, sigmoidalnetworks are inherently non-local by demonstrating that given a su ciently large number of adjustable weights, single-hidden-layer, sigmoidal MLPs exist that are arbitrarily local and retain the abilityto approximateany continuous function on a compact domain. Partially s...

• Type: Journal
• Year: 1998
• Where: TNN
• Authors: S. Weaver, L. Baird, Marios M. Polycarpou
• Keywords: Local Networks, Neural Networks, Quadratic Cost Function, TNN 1998