University of Cambridge > > Machine Learning @ CUED > A Tutorial on Probabilistic Programming

A Tutorial on Probabilistic Programming

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Zoubin Ghahramani.

In probabilistic programming generative processes are represented via computer programs with internal random choices. Observed data condition execution paths of probabilistic programs. Running a probabilistic program characterizes the posterior distribution of internal random choices and memory state (execution paths) arising from program executions that could have generated the observations. This subsumes Bayesian inference in probabilistic models and so it could be claimed that probabilistic programming is the natural generalization of Bayesian probabilistic modeling.

Probabilistic programming has the potential to greatly reduce the technical and cognitive overhead for writing and designing novel probability models in all quantitative fields. In most probabilistic programming systems, including those that will be introduced in this tutorial (Anglican and Probabilistic-C), inference typically is both decoupled from modeling and fully-automated: probabilistic programs (models) can be written without having to derive and program custom inference algorithms. The space of models specifiable as probabilistic programs is large, allowing probabilistic programming practitioners to easily write richly expressive models that would otherwise be difficult to even mathematically or graphically denote.

This talk is part of the Machine Learning @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity