SPLASH 2021
Sun 17 - Fri 22 October 2021 Chicago, Illinois, United States
Fri 22 Oct 2021 11:20 - 11:35 at Zurich F - PLDI 2021 Papers 4 Chair(s): Baris Kasikci

We present the Sum-Product Probabilistic Language (SPPL), a new system that automatically delivers exact solutions to a broad range of probabilistic inference queries. SPPL uses a new class of symbolic expressions to represent the distribution on execution traces of a probabilistic program that generalize sum-product networks by handling mixed-type distributions, numeric transformations, logical formulas, and pointwise and set-valued constraints. We formalize SPPL in terms of a novel translation strategy from probabilistic programs to sum-product expressions and present new and sound algorithms for exactly conditioning on and computing probabilities of events. We present new techniques for improving the scalability of translation and inference by automatically exploiting conditional independences and repeated structure in SPPL programs. We implement a prototype of SPPL with a modular architecture and evaluate it on a suite of benchmarks that the system is designed to solve, which establish that SPPL is up to 3500x faster than state-of-the-art systems for fairness verification; up to 1000x faster than state-of-the-art symbolic algebra techniques; and can compute exact probabilities of rare events in milliseconds.

Fri 22 Oct

Displayed time zone: Central Time (US & Canada) change

10:50 - 12:10
PLDI 2021 Papers 4SIGPLAN Papers at Zurich F
Chair(s): Baris Kasikci University of Michigan, USA
10:50
15m
Talk
Compiling Stan to Generative Probabilistic Languages and Extension to Deep Probabilistic Programming
SIGPLAN Papers
Guillaume Baudart Inria; ENS; PSL University, Javier Burroni , Martin Hirzel IBM Research, Louis Mandel IBM Research, Avraham Shinnar IBM Research
11:05
15m
Talk
On Probabilistic Termination of Functional Programs with Continuous Distributions
SIGPLAN Papers
Raven Beutner University of Oxford, C.-H. Luke Ong University of Oxford
11:20
15m
Talk
SPPL: A Probabilistic Programming System with Exact and Scalable Symbolic Inference
SIGPLAN Papers
Feras Saad Massachusetts Institute of Technology, Martin C. Rinard Massachusetts Institute of Technology, Vikash K. Mansinghka MIT
DOI
11:35
15m
Talk
Cyclic Program Synthesis
SIGPLAN Papers
Shachar Itzhaky Technion, Hila Peleg Technion, Nadia Polikarpova University of California at San Diego, Reuben N. S. Rowe University of Kent, Ilya Sergey National University of Singapore
DOI
11:50
20m
Live Q&A
Discussion, Questions and Answers
SIGPLAN Papers