The Complexity of Bayesian Network Learning: Revisiting the Superstructure (Full Version) Anonymous Author(s) Affiliation Address email

We investigate the parameterized complexity of Bayesian Network Structure Learn-1 ing (BNSL), a classical problem that has received significant attention in empirical2 but also purely theoretical studies. We follow up on previous works that have3 analyzed the complexity of BNSL w.r.t. the so-called superstructure of the input.4 While known results imply that BNSL is unlikely to be fixed-parameter tractable5 even when parameterized by the size of a vertex cover in the superstructure, here we6 show that a different kind of parameterization--notably by the size of a feedback7 edge set--yields fixed-parameter tractability. We proceed by showing that this8 result can be strengthened to a localized version of the feedback edge set, and9 provide corresponding lower bounds that complement previous results to provide a10 complexity classification of BNSL w.r.t.