Dipl.-Ing. Martin Trapp, OFAI Wien und TU Graz
The Turing language for probabilistic programming


Probabilistic programming promises to simplify and democratize probabilistic machine learning, but successful probabilistic programming systems require flexible, generic and efficient inference engines. In this talk I will present the probabilistic programming language called Turing which is under constant development together with the University of Cambridge, the University of Edinburgh and the University of Oxford. Turing has a very simple syntax and makes full use of the numerical capabilities in the Julia programming language, including all implemented probability distributions and automatic differentiation. Moreover, Turing supports a wide range of popular Monte Carlo algorithms including several Hamiltonian Monte Carlo (HMC) algorithms and various particle MCMC (PMCMC) samplers. Most importantly, Turing inference is composable: it combines MCMC operations on subsets of variables, for example using a combination of an HMC engine and a particle Gibbs (PG) engine.

Time: Monday, 10th of December 2018, 6:30 p.m. sharp

Location: Oesterreichisches Forschungsinstitut fuer Artificial Intelligence (OFAI), Freyung 6, Stiege 6, 1010 Wien