A Comparative Study of Different Neural Networks Learning Algorithms for Forecasting Indian Gold Prices

Abstract: Artificial Neural Networks (ANN) have been widely used for forecasting purposes in various fields of science and engineering, economics, finance etc. This paper studies the Artificial Neural Networks based forecasting of Indian gold prices using three learning algorithms, namely, Standard Back Propagation (SBP), Back Propagation with Bayesian Regularization (BPR) and Levenberg-Marquardt (LM) algorithm. The performance of these three learning algorithms is compared using the statistical measures. The study found that the Levenberg-Marquardt learning algorithm based Artificial Neural Network model outperforms the other two learning algorithms based Artificial Neural Network Models in forecasting the Indian Gold prices.

Keywords: Artificial Neural Networks, Standard Back Propagation, Back Propagation with Bayesian Regularization, Lavenberg-Marquardt algorithm.