bn2cnf is a translator associating with an input graphical model (a Bayesian network or a Markov network) an output which consists of a CNF formula and a weight map. The weight of any piece of evidence for the input model can be computed as the weighted model count of the corresponding output. Several encoding schemes can be considered depending on the options used.

Any category    Compiler   

2016 Anicet Bart, Frédéric Koriche, Jean-Marie Lagniez, Pierre Marquis, An Improved CNF Encoding Scheme for Probabilistic Inference in 22nd European Conference on Artificial Intelligence (ECAI'16), pp. 613-621, 2016.