Amortized Inference for Probabilistic Programs

Abstract

Probabilistic programming holds the promise of revolutionizing computational systems by enabling non-experts to embed sophisticated probabilistic AI: machine learning, natural language processing, and computer vision. Stanford set out to radically accelerate probabilistic programming systems by targeting the full implementation stack from inference algorithms to hardware. They have made significant advances in inference algorithms, compilation techniques, and applications. Many of these advances have been released as open source software and/or transitioned to open source projects carried on by industry partners. This has contributed to major growth in the probabilistic programming community in both academia and industry. They expect in the near future to see further growth and uses in high-value applications across diverse sectors.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 01, 2019
Accession Number
AD1067827

Entities

People

  • Noah Goodman

Organizations

  • Stanford University

Tags

Communities of Interest

  • Advanced Electronics
  • Autonomy
  • Ground and Sea Platforms

DTIC Thesaurus Topics

  • Air Force
  • Algorithms
  • Bayesian Networks
  • Computational Science
  • Computer Languages
  • Computer Programming
  • Computer Programs
  • Computer Vision
  • Computers
  • Data Analysis
  • Language
  • Machine Learning
  • Monte Carlo Method
  • Natural Language Processing
  • Neural Networks
  • Probabilistic Models
  • Web Browsers

Fields of Study

  • Computer science

Readers

  • Computational Linguistics
  • Economics
  • Research Science/Academic Research

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy