On-Line Learning for Multi Layered Neural Network: Application to Time Series Prediction

摘要

In this paper a new procedure to continuously adjust weights in a multi alyered neural networks is proposed. The network is initially trained by using traditional Backpropagation algorithm. After this first step, non linear programming techniques are used in order to propely on line calculate the new weights sets. This methodology is tailored to be used in time varying (or non-stationary) models. eliminating necessity of ratraining. Numerical results for the S&P 500 series are presented.