A characterization of Markov qquivalence classes for directed acyclic graphs with latent variables
Proceedings of the Twenty-Third Conference Conference on Uncertainty in Artificial Intelligence (2007)
Different directed acyclic graphs (DAGs) may be Markov equivalent in the sense that they entail the same conditional indepen- dence relations among the observed variables. Meek (1995) characterizes Markov equiva- lence classes for DAGs (with no latent vari- ables) by presenting a set of orientation rules that can correctly identify all arrow orienta- tions shared by all DAGs in a Markov equiv- alence class, given a member of that class. For DAG models with latent variables, maxi- mal ancestral graphs (MAGs) provide a neat representation that facilitates model search. Earlier work (Ali et al. 2005) has identified a set of orientation rules sufficient to con- struct all arrowheads common to a Markov equivalence class of MAGs. In this paper, we provide extra rules sufficient to construct all common tails as well. We end up with a set of orientation rules sound and complete for identifying commonalities across a Markov equivalence class of MAGs, which is partic- ularly useful for causal inference.
Copyright © UAI 2007, AUAI Press.
Access to external full text or publisher's version may require subscription.
ISBN of the source publication: 0974903930
Zhang, J. (2007). A characterization of Markov qquivalence classes for directed acyclic graphs with latent variables. In R. Parr & L. van der Gaag (Eds.), Proceedings of the Twenty-Third Conference Conference on Uncertainty in Artificial Intelligence (2007) (pp.450-457). Corvallis, Oregon: AUAI Press. Retrieved from https://dslpitt.org/uai/papers/07/p450-zhang.pdf