We apply recent methods in stochastic data analysis for discovering a set offew stochastic variables that represent the relevant information on amultivariate stochastic system, used as input for artificial neural networksmodels for air quality forecast. We show that using these derived variables asinput variables for training the neural networks it is possible tosignificantly reduce the amount of input variables necessary for the neuralnetwork model, without considerably changing the predictive power of the model.The reduced set of variables including these derived variables is thereforeproposed as optimal variable set for training neural networks models inforecasting geophysical and weather properties. Finally, we briefly discussother possible applications of such optimized neural network models.
展开▼
