Inference Compilation and Universal Probabilistic Programming

Tuan Anh Le, Atilim Gunes Baydin, Frank Wood
Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, PMLR 54:1338-1348, 2017.

Abstract

We introduce a method for using deep neural networks to amortize the cost of inference in models from the family induced by universal probabilistic programming languages, establishing a framework that combines the strengths of probabilistic programming and deep learning methods. We call what we do “compilation of inference” because our method transforms a denotational specification of an inference problem in the form of a probabilistic program written in a universal programming language into a trained neural network denoted in a neural network specification language. When at test time this neural network is fed observational data and executed, it performs approximate inference in the original model specified by the probabilistic program. Our training objective and learning procedure are designed to allow the trained neural network to be used as a proposal distribution in a sequential importance sampling inference engine. We illustrate our method on mixture models and Captcha solving and show significant speedups in the efficiency of inference.

Cite this Paper

BibTeX
@InProceedings{pmlr-v54-le17a, title = {{Inference Compilation and Universal Probabilistic Programming}}, author = {Le, Tuan Anh and Baydin, Atilim Gunes and Wood, Frank}, booktitle = {Proceedings of the 20th International Conference on Artificial Intelligence and Statistics}, pages = {1338--1348}, year = {2017}, editor = {Singh, Aarti and Zhu, Jerry}, volume = {54}, series = {Proceedings of Machine Learning Research}, month = {20--22 Apr}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v54/le17a/le17a.pdf}, url = {https://proceedings.mlr.press/v54/le17a.html}, abstract = {We introduce a method for using deep neural networks to amortize the cost of inference in models from the family induced by universal probabilistic programming languages, establishing a framework that combines the strengths of probabilistic programming and deep learning methods. We call what we do “compilation of inference” because our method transforms a denotational specification of an inference problem in the form of a probabilistic program written in a universal programming language into a trained neural network denoted in a neural network specification language. When at test time this neural network is fed observational data and executed, it performs approximate inference in the original model specified by the probabilistic program. Our training objective and learning procedure are designed to allow the trained neural network to be used as a proposal distribution in a sequential importance sampling inference engine. We illustrate our method on mixture models and Captcha solving and show significant speedups in the efficiency of inference.} }
Endnote
%0 Conference Paper %T Inference Compilation and Universal Probabilistic Programming %A Tuan Anh Le %A Atilim Gunes Baydin %A Frank Wood %B Proceedings of the 20th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2017 %E Aarti Singh %E Jerry Zhu %F pmlr-v54-le17a %I PMLR %P 1338--1348 %U https://proceedings.mlr.press/v54/le17a.html %V 54 %X We introduce a method for using deep neural networks to amortize the cost of inference in models from the family induced by universal probabilistic programming languages, establishing a framework that combines the strengths of probabilistic programming and deep learning methods. We call what we do “compilation of inference” because our method transforms a denotational specification of an inference problem in the form of a probabilistic program written in a universal programming language into a trained neural network denoted in a neural network specification language. When at test time this neural network is fed observational data and executed, it performs approximate inference in the original model specified by the probabilistic program. Our training objective and learning procedure are designed to allow the trained neural network to be used as a proposal distribution in a sequential importance sampling inference engine. We illustrate our method on mixture models and Captcha solving and show significant speedups in the efficiency of inference.
APA
Le, T.A., Baydin, A.G. & Wood, F.. (2017). Inference Compilation and Universal Probabilistic Programming. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 54:1338-1348 Available from https://proceedings.mlr.press/v54/le17a.html.

Related Material