Quantification of Uncertainty in Mineral Prospectivity Prediction Using Neural Network Ensembles and Interval Neutrosophic Sets

P Kraipeerapun,Chun Che Fung Chun Che Fung,W Brown,Kok Wai Wong Kok Wai Wong

doi:10.1109/ijcnn.2006.247262

P Kraipeerapun, Chun Che Fung Chun Che Fung + Show 2 more

Open Access

https://doi.org/10.1109/ijcnn.2006.247262

Copy DOI

Publication Date: Jan 1, 2006
Citations: 4	License type: cc-by

Affiliation: Murdoch University, University of Western Australia

Abstract

Quantification of uncertainty in mineral prospectivity prediction is an important process to support decision making in mineral exploration, Degree of uncertainty can identify level of quality in the prediction. This paper proposes an approach to predict degrees of favourability for gold deposits together with quantification of uncertainty in the prediction. Geographic Information Systems (GIS) data is applied to the integration of ensemble neural networks and interval neutrosophic sets. Three different neural network architectures are used in this paper. The prediction and its uncertainty are represented in the form of truth-membership, indeterminacy-membership, and false-membership values. Two networks are created for each network architecture to predict degrees of favourability for deposit and non deposit, which are represented by truth and false membership values respectively. Uncertainty or indeterminacy-membership values are estimated from both truth and false membership values. The results obtained using different neural network ensemble techniques are discussed in this paper.

Highlights

Geographic lnfurmatiun Systems (GIS) data is applied to the integratinn of ensemble neural networks and interval neutmsophip w k Three difTewnt neural network architectufis we uscd in this paper
The prcdiction and its unccrtainty arc represented in the form of truth-membership, indeterminacymembership. and false-memkrship values, Two networks are created for sash network architecturn to predict degrees of favr)urahility for deposit and nnn depmit, which are represented by truth add false memhership v.sloes wspectiuely
The results obtained using different neural network ensemhle techniques are diwussed in this paper

Summary

In order to combine the outputs obtained from components

Authorized licensed use limited to: Murdoch University. Downloaded on June 15, 2009 at 03:22 from IEEE Xplore. Of ensemble neural networks, we propose and compare six aggregation techniques which are based on majority vote, averaging, and dynamic averaging techniques. Our proposed techniques have applied the three membership values in the Indeterminacy aggregation instead of the truth-membership only as in most conventional approaches

Falsity BPNN

KPNS GKNS