Prediction of chaotic time series using recurrent neural networks

Abstract
In this paper, we propose to train and use a recurrent artificial neural network(ANN) to predict a chaotic time series. Instead of predicting the next sample in the time series as is normally done, the neural network is trained to produce the sequence that follows a given initial condition. Dynamical parameters from the time series provide the clue in deciding the length of these training sequences. The proposed method has been applied to predicting both periodic and chaotic time series, and is superior to the conventional ANN approach.

This publication has 6 references indexed in Scilit: