An incremental adaptive implementation of functional-link processing for function approximation, time-series prediction, and system identification

This paper presents an adaptive implementation of the functional-link neural network (FLNN) architecture together with a supervised learning algorithm that rapidly determines the weights of the network. The proposed algorithm is able to achieve 'one-shot' training as opposed to iterative training algorithms in the literature. Also discussed is a stepwise updating algorithm that updates the weights of the network while importing new observations. The proposed algorithms have also been tested on several data sets and the simulation shows a very promising result. (C) 1998 Elsevier Science B.V. All rights reserved.