University of Cambridge > > Machine Learning @ CUED > Anglican; Particle MCMC inference for Probabilistic Programs

Anglican; Particle MCMC inference for Probabilistic Programs

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

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

Probabilistic programming languages hold the promise of dramatically accelerating the development of both new statistical models and inference strategies. In probabilistic programs, variables can take on random values at run time and inference is performed by calculating expectation values over all execution traces that are in agreement with a set of observed data. This allows statistical models to be represented in a concise and intuitive manner, enabling more rapid iteration over model variants. Inference schemes, when implemented as a backend to a programming framework, can easily be tested on a large collection of models, enabling a much more systematic comparison of the efficacy of inference strategies.

We introduce Anglican, a probabilistic language that uses particle Markov chain Monte Carlo to perform inference. Our approach is simple to implement, easy to parallelize, and supports accurate inference in models that make use of complex control flow, including stochastic recursion. It also includes primitives from Bayesian nonparametric statistics. Our experiments show that this approach can be more efficient than previously introduced single-site Metropolis-Hastings methods.

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