You are all cordially invited to the next AMLab colloquium next Tuesday, November 17 at 16:00 in C3.163, where Antti Hyttinen from the Helsinki Institute for Information Technology and the University of Helsinki will give a talk titled “Constraint Satisfaction Approach to Causal Inference”.
Abstract: A causal model and the accompanied causal graph explain how a system behaves when interventions are applied to it. In constraint-based causal discovery, statistical independence and dependence relations testable in the observed data are used to reason about the properties of the causal graph. The presentation will outline how constraint-based causal discovery can be achieved using constraint satisfaction. This includes encodings of the d-separation property in propositional logic and solving the constraint satisfaction problem using SAT-solvers, Weighted Partial MaxSAT-solvers and Answer Set Programming solvers. I will also outline how the framework can be used for other problems, such as causal effect estimation.