Probabilistic programming with programmable inference

© 2018 Copyright held by the owner/author(s). We introduce inference metaprogramming for probabilistic programming languages, including new language constructs, a formalism, and the first demonstration of effectiveness in practice. Instead of relying on rigid black-box inference algorithms hard-code...

Full description

Bibliographic Details
Main Authors: Mansinghka, Vikash K., Schaechtle, Ulrich, Handa, Shivam, Radul, Alexey, Chen, Yutian, Rinard, Martin
Other Authors: MIT-IBM Watson AI Lab
Format: Article
Language:English
Published: ACM 2021
Online Access:https://hdl.handle.net/1721.1/136984
_version_ 1826198383185887232
author Mansinghka, Vikash K.
Schaechtle, Ulrich
Handa, Shivam
Radul, Alexey
Chen, Yutian
Rinard, Martin
author2 MIT-IBM Watson AI Lab
author_facet MIT-IBM Watson AI Lab
Mansinghka, Vikash K.
Schaechtle, Ulrich
Handa, Shivam
Radul, Alexey
Chen, Yutian
Rinard, Martin
author_sort Mansinghka, Vikash K.
collection MIT
description © 2018 Copyright held by the owner/author(s). We introduce inference metaprogramming for probabilistic programming languages, including new language constructs, a formalism, and the first demonstration of effectiveness in practice. Instead of relying on rigid black-box inference algorithms hard-coded into the language implementation as in previous probabilistic programming languages, inference metaprogramming enables developers to 1) dynamically decompose inference problems into subproblems, 2) apply inference tactics to subproblems, 3) alternate between incorporating new data and performing inference over existing data, and 4) explore multiple execution traces of the probabilistic program at once. Implemented tactics include gradient-based optimization, Markov chain Monte Carlo, variational inference, and sequental Monte Carlo techniques. Inference metaprogramming enables the concise expression of probabilistic models and inference algorithms across diverse fields, such as computer vision, data science, and robotics, within a single probabilistic programming language.
first_indexed 2024-09-23T11:04:00Z
format Article
id mit-1721.1/136984
institution Massachusetts Institute of Technology
language English
last_indexed 2024-09-23T11:04:00Z
publishDate 2021
publisher ACM
record_format dspace
spelling mit-1721.1/1369842023-04-10T20:10:40Z Probabilistic programming with programmable inference Mansinghka, Vikash K. Schaechtle, Ulrich Handa, Shivam Radul, Alexey Chen, Yutian Rinard, Martin MIT-IBM Watson AI Lab Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory © 2018 Copyright held by the owner/author(s). We introduce inference metaprogramming for probabilistic programming languages, including new language constructs, a formalism, and the first demonstration of effectiveness in practice. Instead of relying on rigid black-box inference algorithms hard-coded into the language implementation as in previous probabilistic programming languages, inference metaprogramming enables developers to 1) dynamically decompose inference problems into subproblems, 2) apply inference tactics to subproblems, 3) alternate between incorporating new data and performing inference over existing data, and 4) explore multiple execution traces of the probabilistic program at once. Implemented tactics include gradient-based optimization, Markov chain Monte Carlo, variational inference, and sequental Monte Carlo techniques. Inference metaprogramming enables the concise expression of probabilistic models and inference algorithms across diverse fields, such as computer vision, data science, and robotics, within a single probabilistic programming language. 2021-11-01T16:59:51Z 2021-11-01T16:59:51Z 2018-06-11 2019-07-02T16:39:26Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/136984 Mansinghka, Vikash K., Schaechtle, Ulrich, Handa, Shivam, Radul, Alexey, Chen, Yutian et al. 2018. "Probabilistic programming with programmable inference." en 10.1145/3192366.3192409 Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf ACM Other repository
spellingShingle Mansinghka, Vikash K.
Schaechtle, Ulrich
Handa, Shivam
Radul, Alexey
Chen, Yutian
Rinard, Martin
Probabilistic programming with programmable inference
title Probabilistic programming with programmable inference
title_full Probabilistic programming with programmable inference
title_fullStr Probabilistic programming with programmable inference
title_full_unstemmed Probabilistic programming with programmable inference
title_short Probabilistic programming with programmable inference
title_sort probabilistic programming with programmable inference
url https://hdl.handle.net/1721.1/136984
work_keys_str_mv AT mansinghkavikashk probabilisticprogrammingwithprogrammableinference
AT schaechtleulrich probabilisticprogrammingwithprogrammableinference
AT handashivam probabilisticprogrammingwithprogrammableinference
AT radulalexey probabilisticprogrammingwithprogrammableinference
AT chenyutian probabilisticprogrammingwithprogrammableinference
AT rinardmartin probabilisticprogrammingwithprogrammableinference