WP4: Data Integration and ML

Contributers: AAU, AU, I•GIS

Data integration and machine learning Following the DSM approach, an advanced ML technique. By using CNN, a powerful new ML method, we expect a much higher prediction accuracy than classical ML techniques. As CNNs require the use of images as input, the gathered data will be integrated to allow image extraction, building on ongoing work. The project will not only produce predictive maps for the target variables, but also uncertainty maps and model interpretation, which both constitute significant gaps within current soil mapping studies. Since the predictive maps and their associated uncertainties will be used by decision-makers, it is crucial to get some insight into the developed CNN models drawing on recent work on explainability.