Causal Structure Learning

TLDR

Graphical models represent multivariate distributions as graphs, and causal models—a special class—also encode distributions under interventions, enabling predictive decisions but requiring underlying assumptions for learning. The study discusses several recently proposed structure learning algorithms and their assumptions. The authors compare the empirical performance of these algorithms under various scenarios.

Abstract

Graphical models can represent a multivariate distribution in a convenient and accessible form as a graph. Causal models can be viewed as a special class of graphical models that represent not only the distribution of the observed system but also the distributions under external interventions. They hence enable predictions under hypothetical interventions, which is important for decision making. The challenging task of learning causal models from data always relies on some underlying assumptions. We discuss several recently proposed structure learning algorithms and their assumptions, and we compare their empirical performance under various scenarios.