Generating Random Bayesian networks with constraints on induced width
Jaime S. Ide, Fábio Gagliardi Cozman, Fabio Tozeto Ramos · 2004
Abstract. We present algorithms for the generation of uniformly distributed Bayesian networks with constraints on induced width. The algorithms use ergodic Markov chains to generate samples. The introduction of constraints on induced width leads to realistic networks but requires new techniques. A tool that generates random networks is presented and applications are discussed. 1