FEED-FORWARD NEURAL NETWORKS REGRESSION ANALYSIS WITH GENETIC ALGORITHMS: APPLICATIONS IN ECONOMICS AND FINANCE

In this paper feed-forward neural networks are examined using genetic algorithms in the training process instead of error backpropagation algorithm. Additionally, real encoding is preferred to binary encoding as it is more appropriate to find the optimum weights. Learning and momentum rates are used for the weight updating as in the case of the error backpropagation algorithm. Some empirical examples as well as the programming routines in MATLAB are provided in the paper