Tractable Probabilistic Circuits

Explore recent developments in tractable probabilistic circuit models in this 37-minute talk by Guy Van den Broeck from UCLA. Gain insights into non-causal probabilistic models that efficiently compute properties like marginal probabilities, entropies, expectations, and other relevant queries. Discover how these models are effectively learned from data and their potential to provide an efficient probabilistic foundation for causal inference algorithms. Delve into the algorithmic aspects of causal inference and understand the implications of these advancements in the field of probabilistic modeling.

Tractable Probabilistic Circuits