Swift: Compiled Inference for Probabilistic Programming Languages
Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence. Pages 3637-3644.
Abstract
A probabilistic program defines a probability measure over its semantic structures. One common goal of probabilistic programming languages (PPLs) is to compute posterior probabilities for arbitrary models and queries, given observed evidence, using a generic inference engine. Most PPL inference engines — even the compiled ones — incur significant runtime interpretation overhead, especially for contingent and open-universe models. This paper describes Swift, a compiler for the BLOG PPL. Swift-generated code incorporates optimizations that eliminate interpretation overhead, maintain dynamic dependencies efficiently, and handle memory management for possible worlds of varying sizes. Experiments comparing Swift with other PPL engines on avariety of inference problems demonstrate speedups ranging from 12x to326x.
@inproceedings{ijcai2016p512,
title = {Swift: Compiled Inference for Probabilistic Programming Languages},
author = {Wu, Yi and Li, Lei and Russell, Stuart J. and Bodík, Rastislav},
booktitle = {Proceedings of the Twenty-Fifth International Joint Conference on
Artificial Intelligence, {IJCAI-16}},
publisher = {AAAI Press / International Joint Conferences on Artificial Intelligence},
editor = {Subbarao Kambhampati},
pages = {3637--3644},
year = {2016},
month = {7},
url = {https://ijcai.org/proceedings/2016/512},
}
TY - CPAPER TI - Swift: Compiled Inference for Probabilistic Programming Languages AU - Wu, Yi AU - Li, Lei AU - Russell, Stuart J. AU - Bodík, Rastislav T2 - Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-16 PB - AAAI Press / International Joint Conferences on Artificial Intelligence PY - 2016 SP - 3637 EP - 3644 UR - https://ijcai.org/proceedings/2016/512 ER -Copied
Machine-readable versions of this paper, for tools and AI agents:
