Stochastic Logic Programs: Sampling, Inference and Applications
This work addresses inference challenges in probabilistic logic programming for researchers in AI and machine learning, presenting incremental improvements by adapting existing statistical methods to SLPs.
The paper tackles inference in stochastic logic programs (SLPs) by developing exact and approximate algorithms based on variable elimination and importance sampling, and applies them to represent prior distributions in machine learning, such as for logic programs and Bayes net structures, using Metropolis-Hastings for posterior sampling with a Prolog implementation.
Algorithms for exact and approximate inference in stochastic logic programs (SLPs) are presented, based respectively, on variable elimination and importance sampling. We then show how SLPs can be used to represent prior distributions for machine learning, using (i) logic programs and (ii) Bayes net structures as examples. Drawing on existing work in statistics, we apply the Metropolis-Hasting algorithm to construct a Markov chain which samples from the posterior distribution. A Prolog implementation for this is described. We also discuss the possibility of constructing explicit representations of the posterior.