Application of machine learning algorithms to mineral prospectivity mapping

In the modern era of diminishing returns on fixed exploration budgets, challenging targets, and ever-increasing numbers of multi-parameter datasets, proper management and integration of available data is a crucial component of any mineral exploration program. Machine learning algorithms have successfully been used for years by the technology sector to accomplish just this task on their databases, and recent developments aim at appropriating these successes to the field of mineral exploration. Framing the exploration task as a supervised learning problem, the geological, geochemical and geophysical information can be used as training data, and known mineral occurences can be used as training labels. The goal is to parameterize the complex relationships between the data and the labels such that mineral potential can be estimated in under-explored regions using available geoscience data. Numerous models and algorithms have been attempted for mineral prospectivity mapping in the past, and in this thesis we propose two new approaches. The first is a modified support vector machine algorithm which incorporates uncertainties on both the data and the labels. Due to the nature of geoscience data and the characteristics of the mineral prospectivity mapping problem, uncertainties are known to be very important. The algorithm is demonstrated on a synthetic dataset to highlight this importance, and then used to generate a prospectivity map for copper-gold porphyry targets in central British Columbia using the QUEST dataset as a case study. The second approach, convolutional neural networks, was selected due to its inherent sensitivity to spatial patterns. Though neural networks have been used for mineral prospectivity mapping, convolutional neural nets have yet to be applied to the problem. Having gained extreme popularity in the computer vision field for tasks involving image segmentation, identification and anomaly detection, the algorithm is ideally suited to handle the mineral prospectivity mapping problem. A CNN code is developed in Julia, then tested on a synthetic example to illustrate its effectiveness at identifying coincident structures in a multi-modal dataset. Finally, a subset of the QUEST dataset is used to generate a prospectivity map using CNNs.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.