Counting and exploring sizes of Markov equivalence classes of directed acyclic graphs

When learning a directed acyclic graph (DAG) model via observational data, one generally cannot identify the underlying DAG, but can potentially obtain a Markov equivalence class. The size (the number of DAGs) of a Markov equivalence class is crucial to infer causal effects or to learn the exact causal DAG via further interventions. Given a set of Markov equivalence classes, the distribution of their sizes is a key consideration in developing learning methods. However, counting the size of an equivalence class with many vertices is usually computationally infeasible, and the existing literature reports the size distributions only for equivalence classes with ten or fewer vertices. In this paper, we develop a method to compute the size of a Markov equivalence class. We first show that there are five types of Markov equivalence classes whose sizes can be formulated as five functions of the number of vertices respectively. Then we introduce a new concept of a rooted sub-class. The graph representations of rooted subclasses of a Markov equivalence class are used to partition this class recursively until the sizes of all rooted subclasses can be computed via the five functions. The proposed size counting is efficient for Markov equivalence classes of sparse DAGs with hundreds of vertices. Finally, we explore the size and edge distributions of Markov equivalence classes and find experimentally that, in general, (1) most Markov equivalence classes are half completed and their average sizes are small, and (2) the sizes of sparse classes grow approximately exponentially with the numbers of vertices.

[1]  Mathias Drton,et al.  Robust graphical modeling of gene networks using classical and alternative t-distributions , 2010, 1009.3669.

[2]  Martin A. Nowak,et al.  Inferring Cellular Networks Using Probabilistic Graphical Models , 2004 .

[3]  P. Spirtes,et al.  Causation, prediction, and search , 1993 .

[4]  Robert Castelo,et al.  Learning Essential Graph Markov Models From Data , 2002, Probabilistic Graphical Models.

[5]  David Maxwell Chickering,et al.  Learning Equivalence Classes of Bayesian Network Structures , 1996, UAI.

[6]  Michael D. Perlman,et al.  The size distribution for Markov equivalence classes of acyclic digraph models , 2002, Artif. Intell..

[7]  D. Madigan,et al.  A characterization of Markov equivalence classes for acyclic digraphs , 1997 .

[8]  M. Gerstein,et al.  A Bayesian Networks Approach for Predicting Protein-Protein Interactions from Genomic Data , 2003, Science.

[9]  Robert E. Tarjan,et al.  Simple Linear-Time Algorithms to Test Chordality of Graphs, Test Acyclicity of Hypergraphs, and Selectively Reduce Acyclic Hypergraphs , 1984, SIAM J. Comput..

[10]  Bin Yu,et al.  Supplement to "Reversible MCMC on Markov equivalence classes of sparse directed acyclic graphs" , 2013 .

[11]  Bin Yu,et al.  Reversible MCMC on Markov equivalence classes of sparse directed acyclic graphs , 2012, ArXiv.

[12]  Judea Pearl,et al.  Equivalence and Synthesis of Causal Models , 1990, UAI.

[13]  Yangbo He,et al.  Active Learning of Causal Networks with Intervention Experiments and Optimal Designs , 2008 .

[14]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[15]  M. Maathuis,et al.  Estimating high-dimensional intervention effects from observational data , 2008, 0810.4214.

[16]  D. Heckerman,et al.  A Bayesian Approach to Causal Discovery , 2006 .

[17]  Christopher Meek,et al.  Causal inference and causal explanation with background knowledge , 1995, UAI.

[18]  Bernard Manderick,et al.  Learning Causal Bayesian Networks from Observations and Experiments: A Decision Theoretic Approach , 2006, MDAI.