This area scaled only to small problems in practice, or failed to provide The problems, interest has focused on solving them approximately. Due to the inherent complexity of the exact versions of Two natural problems are weighted model counting and distribution-aware Given a CNF formula and a weight for each assignment of values to variables, Probabilistic logic program from interpretations. Programming and that it is indeed possible to learn the parameters of a Inference algorithms improve upon the state-of-the-art in probabilistic logic The proposed approach is experimentally evaluated.

The algorithm employs Expectation Maximization, and is built on top of The second contribution is anĪlgorithm for parameter estimation in the learning from interpretations Model and knowledge compilation literature. Which can be solved using state-of-the-art methods known from the graphical The inference tasks to well-studied tasks such as weighted model counting, Queries and evidence to a weighted Boolean formula. It is based on a conversion of the program and the The first contribution of this paper is a suite of efficient algorithms for Not really been addressed for probabilistic logic programs before. The marginals given evidence and learning from (partial) interpretations have Tackled for probabilistic logic programs. Inference and learning tasks known from the graphical model community can be Probabilistic logic programs are logic programs in which some of the factsĪre annotated with probabilities. In particular, when the amount of exact inference is equally limited, collapsed compilation is competitive with the state of the art, and outperforms it on several benchmarks. Our experimental evaluation shows that collapsed compilation performs well on standard benchmarks. More- over, by having a partially compiled circuit available during sampling, collapsed compilation has access to a highly effective proposal distribution for importance sampling. These properties are naturally exploited in exact inference, but are difficult to harness for approximate inference. This online collapsing, together with knowledge compilation inference on the remaining variables, naturally exploits local structure and context- specific independence in the distribution. It is a collapsed sampling algorithm that incrementally selects which variable to sample next based on the partial sample obtained so far.

We introduce collapsed compilation, a novel approximate inference algorithm for discrete probabilistic graphical models.

Loopster processing