Drawing conclusions from data is a core task of AI. Inference allows intelligent beings and AIs to make decisions based on observations. For good decisions, possible uncertainties in data and models must be taken into account. Reasoning under uncertainty is an algorithmically challenging problem. The goal of the project is to develop algorithms that make reasoning from large amounts of data and for complex models possible. To this end, a research training group is to be established in which reasoning under uncertainty will be researched.