Research and development
Over years of dedicated research, NGI has developed extensive expertise in applying machine learning (ML) techniques to geotechnical challenges.
Our work spans clustering and anomaly detection, classification tasks, and predictive modelling using advanced approaches such as XGBoost, HDBSCAN, artificial neural networks (ANNs), and convolutional neural networks (CNNs). These techniques have been successfully implemented in various geotechnical applications, including analysing drilling data, mapping deformation patterns, and assessing landslide risks. Our contributions are widely recognised in leading scientific journals, adding to the growing body of knowledge on ML applications in geotechnics.
Figure 3. Reinforcement learning architecture for strategy development and decision-making in rock tunnel excavation (Erharter et al., 2021)
Figure 4. Clustering rock mass drilling data using representation learning with UMAP and HDBSCAN (Hansen and Aarset, 2024)