Using CNLS-net (Connectionist Normalized Local Spline-network) to predict the Mackey-Glass chaotic time series

We use the Connectionist Normalized Local Spline (CNLS) network to learn the dynamics of the Mackey-Glass time-delay differential equation, for the case {tau} = 30. We show the optimum network operating mode and determine the accuracy and robustness of predictions. We obtain pedictions of varying accuracy using some 2--120 minutes of execution time on a Sun SPARC-1 workstation. CNLS-net is capable of very good performance in predicting the Mackey-Glass time series. 11 refs., 4 figs.