Time-series forecasting using GA-tuned radial basis functions

In this paper we provide a nonlinear auto-regressive (NAR) time-series mode l for forecasting applications. The nonlinearity is introduced by using rad ial basis functions. RBF networks are widely used in time-series analysis. Three main parameter sets are involved in RBF learning process. They are th e centers and widths of the radial functions, and their weights. Although t he selection of the RBF centers and widths is important, most reported rese arch has dealt only with the problem of weight optimization by making assum ptions about the centers and widths. Therefore, there is no guarantee for f inding the global optimum with respect to all sets of parameters. In this p aper we use genetic algorithms (GAs) to simultaneously optimize all of the RBF parameters so that an effective time-series is designed and used for fo recasting. An example is presented with promising results. (C) 2001 Publish ed by Elsevier Science Inc.