The group works on a range of topics in the area of probabilistic and symbolic AI, with particular focus on probabilistic machine learning and graphical models, graph and relational learning, neuro-symbolic learning, and automated planning and (model-based) reinforcement learning. A common theme for the group’s research in these areas is an emphasis on trustworthiness through models and solutions that are interpretable, explainable, safe, and robust. Our research spans foundational theories, methodological and algorithmic developments as well as applications in sustainability (e.g. water management and renewable energy) and bioinformatics.

Key areas of research include:

• Probabilistic Methods and Models: probabilistic graphical models; latent variable models; PAC-Bayes methods; (statistical) relational learning; relational Bayesian networks.  
• Decision making and automated planning:  sequential decision making; heuristic and symbolic search;  Markov decision processes; intelligent problem solving; model reasoning and reformulation 
• Safe Reinforcement Learning: strategy representations; verification of learned strategies; shielded reinforcement learning; shielding and learning for multi-agent systems