Abstract:
3D modelling of geological domains is an essential part of orebody modelling and
resource classification. Conventionally, one applies a wireframing technique, which, in
practice, is quite laborious. An automatic approach for building an implicit geological
model is the application of the Support Vector Machine (SVM) algorithm. However, in
this machine learning problem, the accuracy of models significantly depends on the
selected parameters. Here, the authors introduce another criterion for selecting optimum
parameters, in which the indicator variogram is considered as an aid. The results can be
used as an instruction for implicit geomodelling, based on the SVM algorithm where one
is dealing with two categories.
Keywords: SVM, Support Vector Machine, indicator variogram, geological modelling,
implicit geomodelling.