Tue 9 Jan 2018 10:15 - 10:17 at Bradbury - POSTER SESSION (14 posters - not talks)

Ideally, a probabilistic programming language should admit a computable semantics. But languages often provide operators that denote uncomputable functions, such as comparison of real numbers. While the use of these uncomputable operators may result in uncomputable programs, a programmer can productively use these operators and still produce a computable program, such as one that compares a real number drawn from a normal distribution with 0.

We propose locale theory, and particularly non-spatial sublocales, as a constructive semantic framework for probabilistic programming. Whereas in measure theory, measurable spaces may not have a smallest probability-1 subspace (for a given probability distribution), some locales have smallest probability-1 sublocales, called random sublocales. Partial functions that almost surely terminate and discontinuous functions that are almost everywhere continuous become terminating and continuous, respectively, when restricted to random sublocales. We present a definition of disintegration and provide an example distribution where in locale theory, a unique continuous disintegration exists using random sublocales, whereas classically the disintegration is discontinuous and is only unique up to null sets.

Conference Day
Tue 9 Jan

Displayed time zone: Tijuana, Baja California change

10:00 - 10:30
POSTER SESSION (14 posters - not talks) PPS at Bradbury
10:00
2m
Talk
Probabilistic Programming for Robotics
PPS
Nils NappSUNY at Buffalo, Marco GaboardiUniversity at Buffalo, SUNY
10:02
2m
Talk
Game Semantics for Probabilistic Programs
PPS
C.-H. Luke OngUniversity of Oxford, Matthijs VákárUniversity of Oxford
10:04
2m
Talk
Interactive Writing and Debugging of Bayesian Probabilistic Programs
PPS
Javier Burroni, Arjun GuhaUniversity of Massachusetts, Amherst, David JensenUniversity of Massachusetts Amherst
Pre-print
10:06
2m
Talk
Deep Amortized Inference for Probabilistic Programs using Adversarial Compilation
PPS
10:08
2m
Talk
Comparing the speed of probabilistic processes
PPS
Mathias Ruggaard PedersenAalborg University, Nathanaël FijalkowAlan Turing Institute, Giorgio BacciAalborg University, Kim LarsenAalborg University, Radu MardareAalborg University
10:10
2m
Talk
Using Reinforcement Learning for Probabilistic Program Inference
PPS
Avi PfefferCharles River Analytics
10:12
2m
Talk
TensorFlow Distributions
PPS
Link to publication Pre-print
10:15
2m
Talk
Constructive probabilistic semantics with non-spatial locales
PPS
Benjamin ShermanMassachusetts Institute of Technology, USA, Jared TramontanoMassachusetts Institute of Technology, Michael CarbinMIT
Pre-print
10:17
2m
Talk
Using probabilistic programs as proposals
PPS
Marco Cusumano-TownerMIT-CSAIL, Vikash MansinghkaMassachusetts Institute of Technology
10:19
2m
Talk
Probabilistic Program Equivalence for NetKAT
PPS
Steffen SmolkaCornell University, David KahnCornell University, Praveen KumarCornell University, Nate FosterCornell University, Dexter Kozen, Alexandra SilvaUniversity College London
Link to publication File Attached
10:21
2m
Talk
Reasoning about Divergences via Span-liftings
PPS
Tetsuya SatoUniversity at Buffalo, SUNY, USA
10:23
2m
Talk
Probabilistic Models for Assured Position, Navigation and Timing
PPS
Andres Molina-MarkhamThe MITRE Corporation
10:25
2m
Talk
The Support Method of Computing Expectations
PPS
Avi PfefferCharles River Analytics
10:27
2m
Talk
Combining static and dynamic optimizations using closed-form solutions
PPS
Daniel LundénKTH Royal Institute of Technology, David BromanKTH Royal Institute of Technology, Lawrence M. MurrayUppsala University