Learning equivalence classes of bayesian-network structures

Two Bayesian-network structures are said to be equivalent if the set of distributions that can be represented with one of those structures is identical to the set of distributions that ca...