Recent Advances in Counting and Sampling Markov Equivalent DAGs

Marcel Wienöbst, Max Bannach, Maciej Liskiewicz

Abstract: Counting and sampling directed acyclic graphs (DAGs) from a Markov equivalence class are fundamental tasks in graphical causal analysis. In this paper, we discuss recently proposed polynomial-time algorithms for these tasks. The presented result solves a long-standing open problem in graphical modelling. Experiments show that the proposed algorithms are implementable and effective in practice. Our paper is an extended abstract of the work [24], honored as an AAAI-21 Distinguished Paper at the 35th AAAI Conference on Artificial Intelligence.