Put AI to work
June 26-27, 2017: Training
June 27-29, 2017: Tutorials & Conference
New York, NY

Probabilistic programming

9:00am12:30pm Tuesday, June 27, 2017
Implementing AI
Location: Sutton North Level: Advanced
Secondary topics:  Machine Learning
Average rating: *****
(5.00, 4 ratings)

Prerequisite Knowledge

  • A working knowledge of probabilistic data analysis and Python

What you'll learn

  • Understand probabilistic programming, an emerging field that brings together key ideas from probability theory, programming languages, and Turing-universal computation to build AI systems with new capabilities in domains such as computer vision and data science
  • Explore mature or rapidly maturing domain-specific languages for Bayesian statistics and deep learning

Description

Probabilistic inference, a widely used, mathematically rigorous approach for interpreting ambiguous information using models that are uncertain or incomplete, has become central to multiple fields, from big data analytics to robotics and AI to computational modeling of the mind and brain. Unfortunately, it currently requires deep technical expertise. Models and inference algorithms are difficult to communicate, design, implement, validate, and optimize, and inference often appears to be fundamentally intractable.

Probabilistic programming aims to solve these problems by making modeling and inference broadly accessible to nonexperts, especially by facilitating data analysis, enabling experts to tackle problems that are currently infeasible, especially in machine intelligence. Probabilistic programming is based on new formalizations of modeling and inference that bring together key ideas from probability theory, programming languages, and Turing-universal computation.

Vikash Mansinghka surveys the emerging field of probabilistic programming, which aims to make modeling and inference broadly accessible to nonexperts, emphasizing four languages: Stan, for hierarchical Bayesian statistics, applied to data from education and ecology; Edward, for deep learning, applied to handwritten digit recognition; Venture, for structure learning, applied to reimplementing the Automatic Statistician; and BayesDB, for probabilistic data analysis, applied to databases from public health, cognitive development, and neuroinformatics.

Photo of Vikash Mansinghka

Vikash Mansinghka

MIT

Vikash Mansinghka is a research scientist at MIT, where he leads the Probabilistic Computing Project, and a cofounder of Empirical Systems, a new venture-backed AI startup aimed at improving the credibility and transparency of statistical inference. Previously, Vikash cofounded a venture-backed startup based on his research that was acquired by Salesforce, was an advisor to Google DeepMind, and held graduate fellowships at the National Science Foundation and MIT’s Lincoln Laboratory. He served on DARPA’s Information Science and Technology advisory board from 2010 to 2012 and currently serves on the editorial boards for the Journal of Machine Learning Research and Statistics and Computation. Vikash holds a PhD in computation, an MEng in computer science, and BS degrees in mathematics and computer science, all from MIT. His PhD dissertation on natively probabilistic computation won the MIT George M. Sprowls dissertation award in computer science, and his research on the Picture probabilistic programming language won an award at CVPR.

Leave a Comment or Question

Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?

Join the conversation here (requires login)

Comments

Lawrence Davis | DIRECTOR, AI PRACTICE
06/23/2017 12:50pm EDT

I am very interested in the state-of-play and outlook for approaches to making probabilistic programming “broadly accessible to non-experts.”