An Overview of Probabilistic Programming

Explore the principles and applications of probabilistic programming in this comprehensive conference talk from Strange Loop. Delve into three research platforms: BayesDB, Picture, and Venture, each addressing different aspects of probabilistic inference and modeling. Learn how BayesDB enables direct querying of probable implications from data tables, discover Picture's capabilities in 3D scene perception using deep neural networks, and understand Venture's approach to general-purpose probabilistic programming. Gain insights into real-world applications, including Earth satellite database analysis, microbial biomarker assessment, and 3D model inference from single images. Follow along as the speaker, MIT researcher Vikash K. Mansinghka, explains key concepts such as Bayesian estimates, adhoc inferential queries, computer graphics in probabilistic modeling, and Bayesian optimization. Understand how probabilistic programming is revolutionizing data analysis, model building, and artificial intelligence across various domains.

Introduction
The newcomers reality
The experts reality
The translation problem
Bayesian Estimates
Bayes DB
Capabilities
Adhoc Inferential Queries
Model Building Engine
Metaphor
Data Analysis
Design Map
Computer Graphics
Measures of Uncertainty
Picture
Example Problem
Probabilistic Languages
Bayesian Optimization
Tree Search
Optimization