We are pleased to announce the...
NIPS*2008 Workshop on
PROBABILISTIC PROGRAMMING: universal languages and inference;
systems; and applications
December 13th, 2008 Whistler, Canada
[EMAIL PROTECTED]
http://probabilistic-programming.org/
Organized by
Daniel Roy (MIT) Vikash Mansinghka (MIT)
John Winn (MSR-Cambridge) David McAllester (TTI-Chicago)
Joshua Tenenbaum (MIT)
IMPORTANT DATES
========================================================
- OCT 21 2008 Submissions Due OCT 28 2008 Notifications Sent
- DEC 13 2008 Workshop (small probability of DEC 12th) DEC 14
- 2008 Possible second day (interested? let us know.)
BACKGROUND
=============================================================
Probabilistic graphical models provide a formal lingua franca for
modeling and a common target for efficient inference algorithms.
Their introduction gave rise to an extensive body of work in machine
learning, statistics, robotics, vision, biology, neuroscience, AI and
cognitive science. However, many of the most innovative and useful
probabilistic models published by the NIPS community far outstrip the
representational capacity of graphical models and associated inference
techniques. Models are communicated using a mix of natural language,
pseudo code, and mathematical formulae and solved using special
purpose, one-off inference methods. Rather than precise
specifications suitable for automatic inference, graphical models
typically serve as coarse, high-level descriptions, eliding critical
aspects such as fine-grained independence, abstraction and recursion.
PROBABILISTIC PROGRAMMING LANGUAGES (e.g., IBAL [14,15,30],
Csoft/Infer.NET [8], Church [7], PTP [25], PFP/Haskell [26]; and
logical approaches including BLOG [12], iBLOG [36], PRiSM [19], BLP
[22], SLP [13], Markov Logic [18]) aim to close this representational
gap, unifying general purpose programming with probabilistic modeling;
literally, users specify a probabilistic model in its entirety (e.g.,
by writing code that generates a sample from the joint distribution)
and inference follows automatically given the specification. These
languages provide the full power of modern programming languages for
describing complex distributions, and can enable reuse of libraries of
models, support interactive modeling and formal verification, and
provide a much-needed abstraction barrier to foster generic, efficient
inference in universal model classes.
We believe that the probabilistic programming language approach, which
has been emerging over the last 10 years from a range of diverse
fields including machine learning, computational statistics, systems
biology, probabilistic AI, mathematical logic, theoretical computer
science and programming language theory, has the potential to
fundamentally change the way we understand, design, build, test and
deploy probabilistic systems. A NIPS workshop will be a unique
opportunity for this diverse community to meet, share ideas,
collaborate, and help plot the course of this exciting research area.
TOPICS
=================================================================
One of the central goals of the workshop is to connect the NIPS, UAI
and programming languages communities, introducing PL researchers to a
wealth of critical statistical machine learning problems and yielding
an influx of new mathematical ideas, compilation techniques, and
application problems to NIPS.
Topics of the workshop will include (but are not limited to):
- languages, in particular programming languages, for _specifying_
joint and conditional distributions (i.e., probabilistic models)
- inference algorithms for executing probabilistic programs, exact and
approximate, deterministic and randomized
- killer applications (what can we do now that we could not before)
- learning and using structured representations of uncertain beliefs;
large-scale probabilistic knowledge engineering
- relationships between logical and procedural representations
- how to build useful systems and environments for different user
groups
- trade-offs between the various existing languages and systems for
probabilistic programming
- how programming language research can help us, and vice versa:
automatic differentiation; contracts; partial evaluation;
identifying mathematical relationships between fundamental
statistical notions like exchangeability, conditional independence,
and random measures; and programming language notions like
referential transparency, concurrency, and higher-order procedures;
framing key conjectures and open problems
- computable analysis and probability theory, e.g. finding exact,
effective representations of continuous distributions
- analysis of probabilistic programs and recursive stochastic
processes and mechanical verification of probabilistic models and
algorithms
FORMAT
=================================================================
This one day workshop will consist of invited and contributed talks,
spotlights, poster sessions, and round-table discussions. Due to the
considerable interest we have already received and a widely stated
preference for a highly collaborative workshop environment, we
anticipate spending the bulk of the workshop on poster sessions and
open discussion, with very few talks.
We are also soliciting feedback on whether a second day focused more
on work and fostering collaboration would be of interest to our
attendees, especially those only attending the workshops. If you
would be interested in an extra day (timed so as to allow late
afternoon/evening flights out of Vancouver), please contact us.
CALL FOR CONTRIBUTIONS
=================================================
Authors interested in presenting their work and ideas at the workshop
should send an email with subject "Workshop Submission" to
[EMAIL PROTECTED] and include:
* a title authors and emails abstract (2 pages, NIPS style^, PDF or
* PostScript)
(^) http://nips.cc/PaperInformation/StyleFiles
ORGANIZERS
=============================================================
Organizer Mailing List: [EMAIL PROTECTED]
Primary Contacts:
Daniel Roy ([EMAIL PROTECTED])
Vikash Mansinghka ([EMAIL PROTECTED])
Computer Science and Artificial Intelligence Laboratory, MIT
John Winn
[EMAIL PROTECTED]
Microsoft Research Cambridge
David McAllester
[EMAIL PROTECTED]
Toyota Technology Institute at Chicago
Joshua Tenenbaum
[EMAIL PROTECTED]
Brain and Cognitive Sciences, MIT
FUNDING
================================================================
We are grateful to Microsoft Research Cambridge for generously
offering to help fund the workshop.
REFERENCES
=============================================================
See http://probabilistic-programming.org
_______________________________________________
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
