A new worst-case training algorithm for RBF neural networks

摘要

A novel worst-case learning scheme is developed in this paper for training radial-basis-function (RBF) neural networks. It minimizes the maximum error rather than the average error as in the case of conventional least-squares learning. This scheme is applicable to a variety of practical situations where the nature of the applications demands a worst-case modeling solution. The scheme will be presented along with an illustrative example.