Maximum Likelihood Markov Hypertrees

One popular class of such models are Markov networks, which use an undirected graph to represent dependencies among variables. Markov networks of low tree-width (i.e. having a triangulation with small cliques ) allow efficient computations, and are useful as learned probability models [8]. A well studied case is that in which the dependency structure is known in advance. In this case the underlying graph is built based on prior knowledge, and a maximum likelihood Markov network over this specific graph is sought [5].