Applications of Feed-Forward Neural Networks with Error Backpropagation Algorithm and Non-Linear Methods in MATLAB

Abstract

In this paper we examine and present the methodology of feed-forward neural networks with error backpropagation algorithm and non-linear methods. We test some applications of time-series analysis in economics. The first part is consisted by applications following the traditional approach of neural networks. In the second part we propose a weighted input regression. Additionally, we present full programming routines in MATLAB in order to replicate the results and for further research applications, modifications, expansions and improvements.

Keywords: Feed-Forward Neural Networks, Error Backpropagation Algorithm, Non-Linear Methods, Time-Series, Inflation Rate, Treasury Bills, Forecast, MATLAB

JEL Classification: C22, C45, C53, C63, G10