Write a Blog >>

Workshop on probabilistic programming semantics

Probabilistic programming is the idea of expressing probabilistic models and inference methods as programs and transformations, to ease use and reuse. The recent rise of practical implementations as well as research activity in probabilistic programming has renewed the need for semantics to help us share insights and innovations.

This workshop aims to bring programming-language and machine-learning researchers together to advance the semantic foundations of probabilistic programming. Topics include but are not limited to:

  • the denotational semantics of probabilistic functions, open universe, loops, and conditioning;
  • the operational semantics of sampling, exact inference, and MCMC transitions;
  • axiomatic and equational reasoning;
  • types and polymorphism;
  • and last but not least, how semantics informs any aspect of probabilistic programming, be it design, theory, implementation, or applications.

Discussion guidelines for speakers and other participants

This workshop aims to bring programming-language and machine-learning researchers together. The accepted presentations are gathered at http://pps2016.soic.indiana.edu/. The blog format there invites everyone to ask questions and leave comments before and after the in-person workshop. Or read the abstracts and bring your questions!

To foster collaboration and establish common ground, the posters, the discussion period after each talk, and the breaks are crucial. If you are giving a talk, please abide by the time limit of 20 minutes, and consider encouraging clarification questions during your talk. Each discussion period is 10 minutes—longer than usual, because interaction is key. Please ask questions.

Because probabilistic programming is a research area that bridges multiple communities with different vocabularies, it is especially useful for everyone to ask questions like “What do you mean by X?” and “How is X useful to you?”, where X is a term that occurs in a presentation. Of course, it is also useful for presenters to explain terms proactively.

If you are in the workshop room during a talk, please give the speaker your full attention. Otherwise, please enjoy the posters and snacks in the hallway. We accepted 10 submissions as posters and 10 as talks, not on the basis of reviewer scores but based on which medium we thought would be most effective in conveying the material. So, some highly ranked submissions that are more technical in nature are accepted as posters.

Accepted extended abstracts

Title
A Denotational Semantics of a Probabilistic Stream-Processing Language
PPS
Pre-print
A Lambda-Calculus Foundation for Universal Probabilistic Programming
PPS
Pre-print
All You Need is the Monad... What Monad Was That Again?
PPS
Pre-print
An Application of Computable Distributions to the Semantics of Probabilistic Programs
PPS
Pre-print
An Interface for Black Box Learning in Probabilistic Programs
PPS
Pre-print
Coalgebraic Trace Semantics for Probabilistic Processes: Preliminary Proposal
PPS
Pre-print
eXchangeable Random Primitives
PPS
Pre-print
Finite-depth Higher-order Abstract Syntax Trees for Reasoning about Probabilistic Programs
PPS
Pre-print
Fixed Points for Markov Decision Processes
PPS
Pre-print
Insomnia: Types and Modules for Probabilistic Programming
PPS
Pre-print
Making our Own Luck: A Language for Random Generators
PPS
Pre-print
Models for Probabilistic Programs with an Adversary
PPS
Pre-print
Observation Propagation for Importance Sampling with Likelihood Weighting
PPS
Pre-print
On The Semantic Intricacies of Conditioning
PPS
Pre-print
Parameterized Probability Monad
PPS
Pre-print
Problems of the Lightweight Implementation of Probabilistic Programming
PPS
Pre-print
Reasoning about Probability and Nondeterminism
PPS
Pre-print
Reproducing Kernel Hilbert Space Semantics for Probabilistic Programs
PPS
Pre-print
Semantics of Higher-order Probabilistic Programs
PPS
Pre-print
The Semantics of Figaro, an Embedded Probabilistic Programming Language
PPS
Pre-print

Call for extended abstracts

We expect this workshop to be informal, and our goal is to foster collaboration and establish common ground. Thus, the proceedings will not be a formal or archival publication, and we expect to spend only a portion of the workshop day on traditional research talks. Nevertheless, as a concrete basis for fruitful discussions, we call for extended abstracts describing specific and ideally ongoing work on probabilistic programming semantics.

Extended abstracts are up to 2 pages in PDF format. Please submit them by October 16 using EasyChair: https://easychair.org/conferences/?conf=pps2016

You're viewing the program in a time zone which is different from your device's time zone change time zone

Sat 23 Jan

Displayed time zone: Guadalajara, Mexico City, Monterrey change

09:00 - 10:00
Session 1PPS at Room St Petersburg II
Chair(s): Cameron Freer Gamalon
09:00
20m
Talk
All You Need is the Monad... What Monad Was That Again?
PPS
Pre-print
09:20
10m
Meeting
Discussion 1
PPS

09:30
20m
Talk
An Interface for Black Box Learning in Probabilistic Programs
PPS
Jan-Willem van de Meent University of Oxford, Brooks Paige University of Oxford, David Tolpin University of Oxford, Frank Wood University of Oxford
Pre-print
09:50
10m
Meeting
Discussion 2
PPS

10:30 - 12:15
Session 2PPS at Room St Petersburg II
Chair(s): Chad Scherrer Galois, Inc.
10:30
20m
Talk
Models for Probabilistic Programs with an Adversary
PPS
Robert Rand University of Pennsylvania, Steve Zdancewic University of Pennsylvania
Pre-print
10:50
10m
Meeting
DIscussion 3
PPS

11:00
20m
Talk
The Semantics of Figaro, an Embedded Probabilistic Programming Language
PPS
Avi Pfeffer Charles River Analytics, Brian Ruttenberg Charles River Analytics
Pre-print
11:20
10m
Meeting
Discussion 4
PPS

11:30
45m
Meeting
Implementor Panel: What can semantics do for probabilistic programming and what can probabilistic programming do for semantics?
PPS
Angelika Kimmig KU Leuven, Oleg Kiselyov , Jan-Willem van de Meent University of Oxford, Avi Pfeffer Charles River Analytics, M: Frank Wood University of Oxford
14:00 - 15:30
Session 3PPS at Room St Petersburg II
Chair(s): Mitchell Wand Northeastern University
14:00
20m
Talk
A Lambda-Calculus Foundation for Universal Probabilistic Programming
PPS
Johannes Borgström Uppsala University, Ugo Dal Lago University of Bologna, Andrew D. Gordon Microsoft Research and University of Edinburgh, Marcin Szymczak University of Edinburgh
Pre-print
14:20
10m
Meeting
Discussion 5
PPS

14:30
20m
Talk
Making our Own Luck: A Language for Random Generators
PPS
Leonidas Lampropoulos University of Pennsylvania, Benjamin C. Pierce University of Pennsylvania, Cătălin Hriţcu INRIA Paris, John Hughes Chalmers University of Technology, Zoe Paraskevopoulou Princeton University, Li-yao Xia ENS Paris
Pre-print
14:50
10m
Meeting
Discussion 6
PPS

15:00
20m
Talk
Semantics of Higher-order Probabilistic Programs
PPS
Sam Staton University of Oxford, Hongseok Yang University of Oxford, UK, Chris Heunen University of Edinburgh, Ohad Kammar University of Cambridge, Frank Wood University of Oxford
Pre-print
15:20
10m
Meeting
Discussion 7
PPS

15:30 - 16:30
Poster SessionPPS at Room St Petersburg II
15:30
60m
Meeting
Insomnia: Types and Modules for Probabilistic Programming
PPS
Aleksey Kliger Xamarin, Inc., Sean Stromsten BAE Systems, Inc.
Pre-print
15:30
60m
Meeting
Finite-depth Higher-order Abstract Syntax Trees for Reasoning about Probabilistic Programs
PPS
Theophilos Giannakopoulos BAE Systems, Inc., Mitchell Wand Northeastern University, Andrew Cobb Northeastern University
Pre-print
15:30
60m
Meeting
Coalgebraic Trace Semantics for Probabilistic Processes: Preliminary Proposal
PPS
Larry Moss Indiana University, Chung-chieh Shan Indiana University, Alexandra Silva Radboud University Nijmegen
Pre-print
15:30
60m
Meeting
Reasoning about Probability and Nondeterminism
PPS
Faris Abou-Saleh University of Oxford, Kwok-Ho Cheung University of Oxford, Jeremy Gibbons University of Oxford, UK
Pre-print
15:30
60m
Meeting
Fixed Points for Markov Decision Processes
PPS
Johannes Hölzl Technische Universität München
Pre-print
15:30
60m
Meeting
A Denotational Semantics of a Probabilistic Stream-Processing Language
PPS
Yohei Miyamoto Graduate School of Informatics, Kyoto University, Kohei Suenaga , Koji Nakazawa Graduate School of Information Science, Nagoya University
Pre-print
15:30
60m
Meeting
Observation Propagation for Importance Sampling with Likelihood Weighting
PPS
Ryan Culpepper Northeastern University
Pre-print
15:30
60m
Meeting
Problems of the Lightweight Implementation of Probabilistic Programming
PPS
Pre-print
15:30
60m
Meeting
Parameterized Probability Monad
PPS
Adam Ścibior University of Cambridge, Andrew D. Gordon Microsoft Research and University of Edinburgh
Pre-print
15:30
60m
Meeting
Reproducing Kernel Hilbert Space Semantics for Probabilistic Programs
PPS
Adam Ścibior University of Cambridge, Bernhard Schölkopf MPI Tuebingen
Pre-print
16:30 - 18:00
Session 5PPS at Room St Petersburg II
Chair(s): Chung-chieh Shan Indiana University
16:30
20m
Talk
eXchangeable Random Primitives
PPS
Nathanael L. Ackerman Harvard University, Cameron Freer Gamalon, Daniel Roy University of Toronto
Pre-print
16:50
10m
Meeting
Discussion 8
PPS

17:00
20m
Talk
An Application of Computable Distributions to the Semantics of Probabilistic Programs
PPS
Daniel Huang Harvard University, Greg Morrisett Cornell University
Pre-print
17:20
10m
Meeting
Discussion 9
PPS

17:30
20m
Talk
On The Semantic Intricacies of Conditioning
PPS
Friedrich Gretz RWTH Aachen University, Nils Jansen RWTH Aachen University, Benjamin Lucien Kaminski RWTH Aachen University, Joost-Pieter Katoen RWTH Aachen University, Federico Olmedo RWTH Aachen University
Pre-print
17:50
10m
Meeting
Discussion 10
PPS