Skip to main content
Home

Main navigation

  • Home
  • Series
  • People
  • Depts & Colleges
  • Open Education

Main navigation

  • Home
  • Series
  • People
  • Depts & Colleges
  • Open Education

Symbolic Conditioning of Arrays in Probabilistic Programs

Series
International Conference on Functional Programming 2017
Video Embed
Praveen Narayanan, Indiana University, USA, gives the third presentation in the third panel, Applications, in the ICFP 2017 conference. Co-written by Chung-Chief Shan, Indiana University, USA.
Probabilistic programming systems make machine learning methods modular by automating inference. Recent work by Shan and Ramsey (2017) makes inference itself modular by automating the common component of conditioning. Their work introduces a symbolic program transformation that treats conditioning generally via the measure-theoretic notion of disintegration. This technique, however, is limited to conditioning a single scalar variable. As a step towards tackling realistic machine learning applications, we have extended the disintegration algorithm to symbolically condition arrays in probabilistic programs. The extended algorithm implements lifted disintegration, where repetition is treated symbolically and without unrolling loops. The technique uses a language of index variables for tracking expressions at various array levels. We find that the method works well for arbitrarily-sized arrays of independent random choices, with the conditioning step taking time linear in the number of indices needed to select an element.

More in this series

View Series
International Conference on Functional Programming 2017

A Framework for Adaptive Differential Privacy

Daniel Winograd-Cort University of Pennsylvania, USA, gives the first presentation in the third panel, Applications, in the ICFP 2017 conference. Co-written by Andreas Haeberlen and Aaron Roth, University of Pennsylvania, USA.
Previous
International Conference on Functional Programming 2017

Abstracting Definitional Interpreters

David Darais, University of Maryland, USA, gives the first presentation in the fourth panel, Effects, in the ICFP 2017 conference. Co-written by Nicholas Labich, David Van Horn, Phuc C. Nguyen, University of Maryland, USA.
Next
Licence
Creative Commons Attribution-Non-Commercial-Share Alike 2.0 UK: England & Wales; http://creativecommons.org/licenses/by-nc-sa/2.0/uk/

Episode Information

Series
International Conference on Functional Programming 2017
People
Praveen Narayanan
Keywords
programming
computing
technology
Department: Department of Computer Science
Date Added: 13/12/2017
Duration: 00:17:13

Subscribe

Apple Podcast Video Video RSS Feed

Download

Download Video

Footer

  • About
  • Accessibility
  • Contribute
  • Copyright
  • Contact
  • Privacy
'Oxford Podcasts' Twitter Account @oxfordpodcasts | MediaPub Publishing Portal for Oxford Podcast Contributors | Upcoming Talks in Oxford | © 2011-2022 The University of Oxford