Probabilistic AI and Reasoning

Artificial intelligence and data science deal with a variety of complex problems ranging from inventory management to controlling humanoid robots. For such tasks it is not immediately clear how to manually program good solutions. Instead, many approaches to successfully dealing with such complex problems are based on modelling the problem in a formal framework (such as logic, a graphical model, or Bayesian network), and applying corresponding reasoning or optimisation techniques to find solutions. This course aims to introduce a number of such frameworks and how they can be used to model real-world problems and often also the uncertainties that arise in them. This includes discussing relations between these models and their associated methods like search, inference, learning and optimisation. The specific topics comprise an introduction to AI, heuristic search, logical modelling, constraint satisfaction, graphical models, Bayesian networks, relational probabilisitic models, utility and actions, planning, (partially observable) Markov decision processes, statistical learning, reinforcement learning, and multi-agent decision making (game theory).