Locating centers for radial basis function neural netwroks using
                              multigrid methods

                             Marietta J. Tretter
                         Department Business Analysis
                             Texas A&M University
                        College Station, TX 77843-4217
                            E021MT@tamvm1.tamu.edu


Radial basis function approximation is well known.  This method has gained
some popularity more recently as a neural network technique.  It offers
generally faster convergence than backpropagation neural networks and tends to
produce better results in the context of time series forecasting.  One problem
with radial basis functions is the location of centers for interpolation.
This paper explores the use of multigrid methods in improving radial basis
function neural networks in the context of business time series forecasting.