Zoom Talk: Thomas Icard (Stanford)
Please contact firstname.lastname@example.org for the password.
Logical Foundations of Causal Modeling
The aim of this talk is to convey some of the rich logical structure that can be found in contemporary mathematical and computational models of causality. Specifically, we explore probabilistic (conditional) logical languages interpreted over "structural causal models" -- as well as other types of models such as (causal) Bayesian networks and probabilistic programs -- and present some of the fundamental results about these systems: how they relate in terms of expressive power (encompassing the so called Causal Hierarchy), axiomatization, complexity, and others. We also explain how these foundational issues in logic connect to practical issues in causal inference, reinforcement learning, and other areas of machine learning and data science.