A Language for Counterfactual Generative Models
Zenna Tavares,u00a0James Koppel,u00a0Xin Zhang,u00a0Ria Das,u00a0Armando Solar-Lezama
We present Omega, a probabilistic programming language with support for counterfactual inference. Counterfactual inference means to observe some fact in the present, and infer what would have happened had some past intervention been taken, e.g. u201cgiven that medication was not effective at dose x, what is the probability that it would have been effective at dose 2x?.u201d We accomplish this by introducing a new operator to probabilistic programming akin to Pearlu2019s do, define its formal semantics, provide an implementation, and demonstrate its utility through examples in a variety of simulation models.
