2013/03/27 by Ross D. Shachter, Shachter, Ross D., Mark A. Peot +1
Computer Science · #Artificial Intelligence (cs.AI) #Bayesian Modeling and Causal Inference #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Logic, Reasoning, and Knowledge #Machine Learning and Algorithms
paper · pdf · doi:10.48550/arxiv.1304.1526
openalex publication_date 2013/03/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
A number of algorithms have been developed to solve probabilistic inference\nproblems on belief networks. These algorithms can be divided into two main\ngroups: exact techniques which exploit the conditional independence revealed\nwhen the graph structure is relatively sparse, and probabilistic sampling\ntechniques which exploit the "conductance" of an embedded Markov chain when the\nconditional probabilities have non-extreme values. In this paper, we\ninvestigate a family of "forward" Monte Carlo sampling techniques similar to\nLogic Sampling [Henrion, 1988] which appear to perform well even in some\nmultiply connected networks with extreme conditional probabilities, and thus\nwould be generally applicable. We consider several enhancements which reduce\nthe posterior variance using this approach and propose a framework and criteria\nfor choosing when to use those enhancements.\n