Abstract
Understanding the behaviour of a system under the influence of interventions is the ultimate goal of many scientific studies. While such causal relationships are ideally inferred from interventional data (obtained through controlled experiments), in many applications one has only access to data that has been obtained by passively observing the system. Hence, there has lately been a growing interest in machine learning methods for inferring causal relationships from observational data given certain assumptions. In this talk, I will present the idea of using causal graphical models as the underlying framework for approaching this problem. More specifically, I will focus on a particular type of method that combines structure learning and causal calculus in order to produce causal effect estimates under an unknown causal structure. In particular, I will present some of our recent work where we adopt this approach to a Bayesian setting in order to better account for the uncertainty in the inference procedure.