Machine learning in mineral prospectivity mapping and target generation for critical raw materials

The aim of mineral prospectivity mapping (MPM) is to delineate areas that are favourable for certain mineral deposit types. This can be based on prior knowledge using a so-called empirical or data-driven approach or by translating expert knowledge into a mathematical formula by using a conceptual or knowledge-driven approach. Both approaches can benefit from machine learning methods using advanced computer algorithms that can learn from data. This learning can either be supervised or unsupervised. Geographical information systems (GIS) provide a perfect platform for conducting MPM, as in these systems, we can automate and build complex systems to construct models that can be used to predict where the best exploration terrains are hidden. This paper aims to describe how machine learning methods can be utilized in MPM in various steps. This is demonstrated via examples of several past and ongoing research and innovation projects.