Intro in structural learning and causal discovery


The talk will be about the assumptions that lie under the rug of such seemingly reliable and time-tested methods as experimental interventions (aka RCT/AB, etc.). Next, we are expected to discuss how modern graphical methods from causal inference can be useful for planning experiments. In particular, we will talk about how to account for pre-collected observed data or expertise when planning factor experiments (when the assumption of additivity or linearity of effects is not met).