NIPS*2012 Workshop
From Probabilistic Programming
PROBABILISTIC PROGRAMMING:
Foundations and Applications
December 7-8th, 2012
Lake Tahoe, Nevada
NIPS*2012 Conference
Organizers:
Vikash Mansinghka (MIT)
Daniel Roy (Cambridge)
Noah Goodman (Stanford)
Contents |
IMPORTANT DATES
| . . . . . . . . . . . . . | Oct. 15, 2012 |
| . . . . . . . . . . . . . | Nov. 01, 2012 (Notifications have been sent out. If you have not heard from us, contact us immediately.) |
| . . . . . . . . . . . . . | Nov. 9, 2012 |
| . . . . . . . . . . . . . | Dec. 7–8, 2012 (two days) |
DAY 1 - FRIDAY DECEMBER 7
7:30am | Opening Address by Vikash Mansinghka | ||
7:40am | Invited talk: Modeling human common sense with probabilistic programs Josh Tenenbaum (MIT) | ||
8:30am | Contributed talk: Automated variational inference in probabilistic programming David Wingate and Theo Weber (Lyric Labs) | ||
9:00am | Coffee | ||
9:30am | A tour through the theoretical foundations of probabilistic programming Dan Roy (Cambridge) | ||
10:00am | Poster Spotlight Presentations
| ||
10:30am | Non-skiers: Informal Session for Discussion, Posters, Demos | ||
12:00pm | Group Lunch | ||
3:30pm | BREAKOUT DISCUSSIONS and POSTERS | ||
5:00pm | Coffee | ||
5.30pm | Invited talk: Probabilistic Programming at Microsoft Research Cambridge: Infer.NET for Real-world Applications Chris Bishop (Microsoft Research) | ||
6.20pm | Announcements and details for Group Dinner |
DAY 2 - SATURDAY DECEMBER 8
7:30am | Invited talk: Open-universe probability models: idea, theory, and applications Stuart Russell (UC Berkeley) | ||
8:20am | Contributed talk: Stan, scalable software for Bayesian modeling Matt Hoffman (Adobe), Bob Carpenter, and Andrew Gelman (Columbia) | ||
8:40am | Contributed talk: A short introduction to probabilistic soft logic Angelika Kimmig (KU Leuven), Stephen Bach, Matthias Broecheler, Bert Huang, Lise Getoor (U. Maryland) | ||
9:00am | Coffee | ||
9:30am | What we gain by representing models using code, illustrated in a new, practical Church system Vikash Mansinghka (MIT) | ||
10:00am | Poster Spotlight Presentations
| ||
10:30am | Non-skiers: Informal Session for Discussion, Posters, Demos | ||
12:00pm | Group Lunch | ||
3:30pm | DEMO SESSION and POSTERS
Systems being demoed include...
Contact us if you would like us to add your system/name to the list. | ||
5:00pm | Coffee | ||
5.30pm | The flexible mind: probabilistic programming as a substrate for cognitive modeling Noah Goodman (Stanford) | ||
6.00pm | Contributed talk: A model-learner pattern for Bayesian reasoning Andy Gordon (Microsoft Research), Mihhail Aizatulin (Open U.), Johannes Borgstroem (Uppsala U.), Guillaume Claret, Thore Graepel, Aditya Nori, Sriram Rajamani, Claudio Russo (Microsoft Research) | ||
6.20pm | Contributed talk: Probabilistic computation for information security Piotr Mardziel (U. Maryland) and Kasturi Raghavan (UCLA) | ||
6.40pm | Closing Remarks by Noah Goodman |
MAILING LIST
We encourage you to join the probabilistic-programming mailing list, in order to receive important announcements and participate in related discussions.
OVERVIEW
An intensive, two-day workshop on PROBABILISTIC PROGRAMMING, with contributed and invited talks, poster sessions, demos, and discussions.
Probabilistic models and inference algorithms have become standard tools for interpreting ambiguous, noisy data and building systems that learn from their experience. However, even simple probabilistic models can require significant effort and specialized expertise to develop and use, frequently involving custom mathematics, algorithm design and software development. State-of-the-art models from Bayesian statistics, artificial intelligence and cognitive science --- especially those involving distributions over infinite data structures, relational structures, worlds with unknown numbers of objects, rich causal simulations of physics and psychology, and the reasoning processes of other agents --- can be difficult to even specify formally, let alone in a machine-executable fashion.
PROBABILISTIC PROGRAMMING aims to close this gap, making variations on commonly-used probabilistic models far easier to develop and use, and pointing the way towards entirely new types of models and inference. The central idea is to represent probabilistic models using ideas from programming, including functional, imperative, and logic-based languages. Most probabilistic programming systems represent distributions algorithmically, in terms of a programming language plus primitives for stochastic choice; some even support inference over Turing-universal languages. Compared with representations of models in terms of their graphical-model structure, these representation languages are often significantly more flexible, but still support the development of general-purpose inference algorithms.
The workshop will cover, and welcomes submissions about, all aspects of probabilistic programming. Some questions of particular interest include:
- What real-world problems can be solved with probabilistic programming systems today? How much problem-specific customization/optimization is needed? Where is general-purpose inference effective?
- What does the probabilistic programming perspective, and in particular the representation of probabilistic models and inference procedures as algorithmic processes, reveal about the computability and complexity of Bayesian inference? When can theory guide the design and use of probabilistic programming systems?
- How can we teach people to write probabilistic programs that work well, without having to teach them how to build an inference engine first? What programming styles support tractability of inference?
- How can central ideas from software engineering --- including debuggers, validation tools, style checkers, program analyses, reusable libraries, and profilers --- help probabilistic programmers and modelers? Which of these tools can be built for probabilistic programs, or help us build probabilistic programming systems?
- What new directions in AI, statistics, and cognitive science would be enabled if we could handle models that took hundreds or thousands of lines of probabilistic code to write?
CONFIRMED KEYNOTE SPEAKERS
- Chris Bishop (Microsoft Research; University of Edinburgh)
- Stuart Russell (Berkeley)
- Josh Tenenbaum (MIT)
ORGANIZERS
Vikash Mansinghka is a Fellow of MIT's Intelligence Initiative and the Computer Science and AI Lab (CSAIL). His research is based on an emerging marriage of the abstractions behind software and hardware with stochastic processes and Bayesian inference. His group develops new probabilistic computing systems as well as applications to machine learning, statistics, cognitive science and AI. His work has yielded new probabilistic programming languages and systems, stochastic digital hardware for real-time Bayesian inference, and practical probabilistic database technology (both academic and commercial) that produces statistically reliable predictions from raw, messy, high-dimensional data tables.Daniel Roy is a Newton Fellow of the Royal Society at the University of Cambridge. His research interests lie at the intersection of computer science, statistics and probability theory; through the study of probabilistic programming languages, his work aims to develop computational perspectives on fundamental ideas in probability theory and statistics. In particular, he is interested in the use of recursion to define nonparametric distributions on data structures; representation theorems that connect computability and probabilistic structures; and the complexity of inference.
Noah Goodman is an Assistant Professor of Psychology, Linguistics (by courtesy), and Computer Science (by courtesy) at Stanford University. Goodman studies computational models of cognition, integrating logic and probability. He co-invented the probabilistic programming language Church and continues to develop novel inference techniques and applications. His work in cognitive science spans reasoning with concepts and intuitive theories; natural language semantics and pragmatics; social cognition; and cognitive development, especially the acquisition of abstract knowledge.
The organizer may be contacted via prob-prog-workshop-organizers@mit.edu .
FUNDING
We are grateful to Lyric Labs (part of Analog Devices) for generously offering to help fund the workshop.