Compiling Stan to generative probabilistic languages and extension to deep probabilistic programming - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2021

Compiling Stan to generative probabilistic languages and extension to deep probabilistic programming

Abstract

Stan is a probabilistic programming language that is popular in the statistics community, with a high-level syntax for expressing probabilistic models. Stan differs by nature from generative probabilistic programming languages like Church, Anglican, or Pyro. This paper presents a comprehensive compilation scheme to compile any Stan model to a generative language and proves its correctness. We use our compilation scheme to build two new backends for the Stanc3 compiler targeting Pyro and NumPyro. Experimental results show that the NumPyro backend yields a 2.3x speedup compared to Stan in geometric mean over 26 benchmarks. Building on Pyro we extend Stan with support for explicit variational inference guides and deep probabilistic models. That way, users familiar with Stan get access to new features without having to learn a fundamentally new language.
Fichier principal
Vignette du fichier
pldi21.pdf (2.4 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03401742 , version 1 (25-10-2021)

Identifiers

Cite

Guillaume Baudart, Javier Burroni, Martin Hirzel, Louis Mandel, Avraham Shinnar. Compiling Stan to generative probabilistic languages and extension to deep probabilistic programming. PLDI '21 - 42nd ACM SIGPLAN International Conference on Programming Language Design and Implementation, Jun 2021, Virtual, Canada. pp.497-510, ⟨10.1145/3453483.3454058⟩. ⟨hal-03401742⟩
39 View
120 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More