Sound Abstraction and Decomposition of Probabilistic Programs

Steven Holtzen, Guy Broeck, Todd Millstein
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:1999-2008, 2018.

Abstract

Probabilistic programming languages are a flexible tool for specifying statistical models, but this flexibility comes at the cost of efficient analysis. It is currently difficult to compactly represent the subtle independence properties of a probabilistic program, and exploit independence properties to decompose inference. Classical graphical model abstractions do capture some properties of the underlying distribution, enabling inference algorithms to operate at the level of the graph topology. However, we observe that graph-based abstractions are often too coarse to capture interesting properties of programs. We propose a form of sound abstraction for probabilistic programs wherein the abstractions are themselves simplified programs. We provide a theoretical foundation for these abstractions, as well as an algorithm to generate them. Experimentally, we also illustrate the practical benefits of our framework as a tool to decompose probabilistic program inference.

Cite this Paper


BibTeX
@InProceedings{pmlr-v80-holtzen18a, title = {Sound Abstraction and Decomposition of Probabilistic Programs}, author = {Holtzen, Steven and Van den Broeck, Guy and Millstein, Todd}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {1999--2008}, year = {2018}, editor = {Dy, Jennifer and Krause, Andreas}, volume = {80}, series = {Proceedings of Machine Learning Research}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/holtzen18a/holtzen18a.pdf}, url = {https://proceedings.mlr.press/v80/holtzen18a.html}, abstract = {Probabilistic programming languages are a flexible tool for specifying statistical models, but this flexibility comes at the cost of efficient analysis. It is currently difficult to compactly represent the subtle independence properties of a probabilistic program, and exploit independence properties to decompose inference. Classical graphical model abstractions do capture some properties of the underlying distribution, enabling inference algorithms to operate at the level of the graph topology. However, we observe that graph-based abstractions are often too coarse to capture interesting properties of programs. We propose a form of sound abstraction for probabilistic programs wherein the abstractions are themselves simplified programs. We provide a theoretical foundation for these abstractions, as well as an algorithm to generate them. Experimentally, we also illustrate the practical benefits of our framework as a tool to decompose probabilistic program inference.} }
Endnote
%0 Conference Paper %T Sound Abstraction and Decomposition of Probabilistic Programs %A Steven Holtzen %A Guy Broeck %A Todd Millstein %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-holtzen18a %I PMLR %P 1999--2008 %U https://proceedings.mlr.press/v80/holtzen18a.html %V 80 %X Probabilistic programming languages are a flexible tool for specifying statistical models, but this flexibility comes at the cost of efficient analysis. It is currently difficult to compactly represent the subtle independence properties of a probabilistic program, and exploit independence properties to decompose inference. Classical graphical model abstractions do capture some properties of the underlying distribution, enabling inference algorithms to operate at the level of the graph topology. However, we observe that graph-based abstractions are often too coarse to capture interesting properties of programs. We propose a form of sound abstraction for probabilistic programs wherein the abstractions are themselves simplified programs. We provide a theoretical foundation for these abstractions, as well as an algorithm to generate them. Experimentally, we also illustrate the practical benefits of our framework as a tool to decompose probabilistic program inference.
APA
Holtzen, S., Broeck, G. & Millstein, T.. (2018). Sound Abstraction and Decomposition of Probabilistic Programs. Proceedings of the 35th International Conference on Machine Learning, in Proceedings of Machine Learning Research 80:1999-2008 Available from https://proceedings.mlr.press/v80/holtzen18a.html.

Related Material