Probabilistic Programming with Densities in SlicStan: Efficient, Flexible and Deterministic
Stan is a probabilistic programming language that has been increasingly used for real-world scalable projects. However, to make practical inference possible, the language sacrifices some of its usability by adopting a block syntax, which lacks compositionality and flexible user-defined functions. Moreover, the semantics of the language has been mainly given in terms of intuition about implementation, and has not been formalised.
This paper provides a formal treatment of the Stan language, and introduces the probabilistic programming language SlicStan — a compositional, self-optimising version of Stan. Our main contributions are: (1) the formalisation of a core subset of Stan through an operational density-based semantics; (2) the design and semantics of the Stan-like language SlicStan, which facilities better code reuse and abstraction through its compositional syntax, more flexible functions, and information-flow type system; and (3) a formal, semantic-preserving procedure for translating SlicStan to Stan.
|Presentation Slides (Probabilistic Programming with Densities in SlicStan.pdf)||1.92MiB|
Wed 16 JanDisplayed time zone: Belfast change
13:45 - 14:51
|Probabilistic Programming with Densities in SlicStan: Efficient, Flexible and Deterministic|
Maria I. Gorinova The University of Edinburgh, Andrew D. Gordon Microsoft Research and University of Edinburgh, Charles Sutton University of EdinburghLink to publication DOI Pre-print Media Attached File Attached
|A Domain Theory for Statistical Probabilistic ProgrammingDistinguished Paper|
Matthijs Vákár University of Oxford, Ohad Kammar University of Edinburgh, Sam Staton University of OxfordLink to publication DOI Pre-print Media Attached File Attached
|Bayesian Synthesis of Probabilistic Programs for Automatic Data Modeling|
Feras Saad Massachusetts Institute of Technology, Marco Cusumano-Towner MIT-CSAIL, Ulrich Schaechtle Massachusetts Institute of Technology, USA, Martin C. Rinard Massachusetts Institute of Technology, Vikash K. Mansinghka MITLink to publication DOI Media Attached File Attached