Inverse problems and imaging have many real-world applications, such as medicine or seismics. Established approaches for their solution are often either model-based or data-based. While model-based approaches are amenable to mathematical analysis, data-based approaches directly relate to the inverse problem at hand and become more useful with the availability of big data sets. This research combines the advantages of both approaches with the aim to develop more accurate, complete and efficient solvers for inverse problems. Particular topics of our research include nonlocal regularization, data assimilation, neural networks, and optimal transport on graphs
People working on this subject within SACS are:
Staff:
Post Doc / PhD
