Vol. 5, No. 1&2 (May 1992) 69-74   

downloaded Downloaded: 58   viewed Viewed: 796

A.Khotanzad and J.H. Lu

 Electrical Engineering Department, Southern Methodist University
Dallas, Texas U.S.A


Abstract    In this paper a nonparametric neural network (NN) technique for prediction of future values of a signal based on its past history is presented. This approach bypasses modeling, identification, and parameter estimation phases that are required by conventional parametric techniques. A multi-layer feed forward NN is employed. It develops an internal model of the signal through a training operation involving past history of the considered signal. Training is performed using the back-propagation algorithm. The trained net is then used to do the forecast. Training is continued during operation to improve performance. The net performance is tested on signals generated by autoregressive (AR) and autoregressive moving average (ARMA) models of different orders and results are compared to optimal forecasts.



1. Khotanzad, A. and Lu, J. H.,”Classification of Invariant Image Representations Using Neural Network , to appear in IEEE Transactions on Acoustics, Speech, and Signal Processing.”,Vol. 38, (1990).

2. Khotanzad, A. and Lu, J. H., “Non-Parametric Prediction of AR Processed Using Neural Networks , Int. conf. Acous., Speech, Signal Proc.”, ICASSP-90, Albuquerque, NM., (April. 1990).

3. Lippmann, R. P.,”An Introduction to Computing with Neural Nets , IEEE ASSP Magazine, Vol. 4, No. 2, (April. 1987), 4-22.

4. Marple, S. L, Digital Spectural Analysis with Applications, Prentice-Hell, New Jersey, (1987).

5. Rumelhart, D. E.and McClelland, J. L. , Parallel Distributed Processing”, Vol. 1, Foundations.MIT press, MA., (1986).

6. Werbos, P.,”Learning How the World Works” Specifications for Predictive Networks in Robots and Brains”, Proceedings of the 1987 IEEE Int. Conf. on SMC”, Vol. 1, (1987).f

International Journal of Engineering
E-mail: office@ije.ir
Web Site: http://www.ije.ir