作者: Eyal Amir , Dan Roth , Rodrigo De Salvo Braz
DOI:
关键词: Variable elimination 、 Theoretical computer science 、 Frequentist inference 、 Graphical model 、 Backward chaining 、 Probabilistic inference 、 Fiducial inference 、 Disjunction introduction 、 Junction tree algorithm 、 Adaptive neuro fuzzy inference system 、 Computer science 、 Algorithm 、 Inference 、 Probabilistic logic network
摘要: Most probabilistic inference algorithms are specified and processed on a propositional level. In the last decade, many proposals for accepting first-order specifications have been presented, but in stage they still operate mostly representation [Poole, 2003] presented method to perform directly level, this is limited special cases. paperwe present first exact algorithm that operates can be applied any model (specified language generalizes undirected graphical models). Our experiments show superior performance comparison with inference.