A neural network nonlinear predictor

Abstract

Summary form only given, as follows. The authors demonstrate that a backpropagation neural network can be used for nonlinear time series prediction. In a computer experiment an example nonlinear time series is used to teach a network the necessary mapping in a supervised manner. Predictor learning curves are presented, showing successful operation. Improvements to the neural network structure with regard to the reduction of the observed performance deficit are discussed.<>