On the use of local search heuristics to improve GES-based Bayesian network learning

摘要: Abstract Bayesian networks learning is computationally expensive even in the case of sacrificing optimality result. Many methods aim at obtaining quality solutions affordable times. Most them are based on local search algorithms, as they allow evaluating candidate a very efficient way, and can be further improved by using search-based metaheuristics to avoid getting stuck optima. This approach has been successfully applied searching for network structures space directed acyclic graphs. Other algorithms equivalence classes. The most important these GES (greedy search). It guarantees optimal under certain conditions. However, it also get optima when from datasets with limited size. article proposes use way improve behaviour such circumstances. These guarantee asymptotical optimality, experiments show that upon score obtained GES.