Machine Learning
Probabilistic reasoning with answer sets
Theory and Practice of Logic Programming
IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence
Hi-index | 0.00 |
I present probabilistic extensions to propositional and first-order logic. The general approach is to define a probability distribution over possible worlds; this distribution is chosen as the unique distribution that agrees with a set of probability constraints over a set of logical sentences and that has maximal entropy. Such distributions are well studied in the Statistics and Machine Learning literature, and using them not only results in simple, intuitive models, but makes available standard inference and learning algorithms.