Probabilistic Programming Concepts

A multitude of different probabilistic programming languages exists today, all extending a traditional programming language with primitives to support modeling of complex, structured probability distributions, advanced inference and learning methods. At the same time, they extend probabilistic graphical models with primitives of programming languages to increase the expressive power of graphical models. I shall provide an overview of the underlying concepts and semantics of these languages as well as sketch their current inference and learning mechanisms. I shall also outline some emerging applications of these languages. During the talk I shall focus on probabilistic extensions of logic programming languages such as Prolog and Datalog, which have been developed since more than 20 years. One advantage of these languages is that they naturally fit both the statistical relational learning / artificial intelligence perspective as they define probability distributions over possible worlds as well as the probabilistic programming language perspective. This talk will be largely based on joint tutorials with Angelika Kimmig and the forthcoming survey paper : Luc De Raedt, Angelika Kimmig, Probabilistic (Logic) Programming Concepts, Machine Learning, in press. An early version of this paper can be found at http://arxiv.org/abs/1312.4328.Invited Tutorialstatus: publishe