PhD Thesis (2022-2025)

Title:  Graph machine learning for geological structural interpretation of sparse observations

Supervisors: Guillaume Caumon (GeoRessources, Université de Lorraine), Radu Stoica (Université de Lorraine, IECL)

The objective of the thesis is to evaluate different graph machine learning models in the context of sparse structural data association. The first step will be to complement or replace existing two-point fault association rules (Godefroy et al., 2019) by substituting geometric criteria by a machine leaning on training models. Then, we will investigate of multi-point rules to account for higher order interactions between data. The proposed approaches will help in the interpretation of faults in sparse data contexts, especially where no 3D seismic image is available.