Augur: a Modeling Language for Data-Parallel Probabilistic Inference
Authors:
Jean-Baptiste Tristan,
Daniel Huang,
Joseph Tassarotti,
Adam Pocock,
Stephen J. Green,
Guy L. Steele Jr
Abstract:
It is time-consuming and error-prone to implement inference procedures for each new probabilistic model. Probabilistic programming addresses this problem by allowing a user to specify the model and having a compiler automatically generate an inference procedure for it. For this approach to be practical, it is important to generate inference code that has reasonable performance. In this paper, we p…
▽ More
It is time-consuming and error-prone to implement inference procedures for each new probabilistic model. Probabilistic programming addresses this problem by allowing a user to specify the model and having a compiler automatically generate an inference procedure for it. For this approach to be practical, it is important to generate inference code that has reasonable performance. In this paper, we present a probabilistic programming language and compiler for Bayesian networks designed to make effective use of data-parallel architectures such as GPUs. Our language is fully integrated within the Scala programming language and benefits from tools such as IDE support, type-checking, and code completion. We show that the compiler can generate data-parallel inference code scalable to thousands of GPU cores by making use of the conditional independence relationships in the Bayesian network.
△ Less
Submitted 10 June, 2014; v1 submitted 12 December, 2013;
originally announced December 2013.
