Comparison of Artificial Neural Networks and a Geostatistical Method in Grade Estimation

In the present paper, the performance of four approaches based on neural networks and geostatistical method for grade estimation are compared and analyzed their performance to find a proper method. Four methods, Multi Layer Perceptron neural network (MLP), Radial Basis Function (RBF), General Regression Neural Network (GRNN) and Ordinary Kriging (OK) are selected for ore grade estimation of variable Fe in an iron deposit located in Iran. For each method, two diagnostic statistics are calculated: Mean square error (MSE) and correlation coefficient. The results show the superiority of the General Regression Neural Network.