Learning and predicting time series by neural networks

Ansgar Freking, Wolfgang Kinzel, Ido Kanter

Artificial neural networks which are trained on a time series are supposed to achieve two abilities: firstly to predict the series many time steps ahead and secondly to learn the r…