Mineral Prospectivity Prediction using Interval Neutrosophic Sets

P. Kraipeerapun, C.C. Fung, W. Brown, and K.-W. Wong (Australia)


interval neutrosophic set, feed-forward backpropagation neural network, GIS, uncertainty, mineral prospectivity mapping


The integration of Geographic Information Systems (GIS) data with neural networks can improve the accuracy of predictive maps showing the favourability for mineral de posits at regional-scale. However, uncertainty is rarely es timated for mineral prospectivity maps. Quantification of uncertainty is very important as it can enhance prediction and supports decision making. This paper proposes an ap proach to the prediction of mineral prospectivity that pro vides an assessment of uncertainty. Interval neutrosophic sets are combined with neural networks to classify map cells as either deposits or barren based on input feature vectors representing exploration data. In this paper, two feed-forward backpropagation neural networks are used to determine truth-membership, indeterminacy-membership, and false-membership values, which together form an in terval neutrosophic set. The INS model improves classi fication performance compared to a threshold method and has the advantage of providing a measure of the uncertainty of the classification.

Important Links:

Go Back