An Approach for Time Series Forecasting by Simulating Stochastic Processes Through Time Lagged Feed-forward Neural Network

In this work an approach for time series forecasting by simulating stochastic processes through time lagged feed-forward neural network is presented. The learning rule used to adjust the neural network (NN) weights is based on the Levenberg-Marquardt method. In function of the long or short term stochastic dependence of the time series, an on-line heuristic law to set the training process and to modify the NN topology is employed. The NN output tends to approximate the current value available from the series by applying a time-delay operator. The approach is tested over samples of the Mackey-Glass delay differential equations (MG). Four sets of parameters for MG definition were used. The performance is shown by forecasting the 18 future values of four time series of 102 data length each. Each time series was simulated by a Monte Carlo of 50 trials at the final of each data series.