[ABNMS-L] [CfP] Fourteenth Annual Bayesian Modeling Applications Workshop (BMAW 2017)

Steven Mascaro steven.mascaro at bayesian-intelligence.com
Wed Jun 28 12:33:12 AEST 2017


The Fourteenth Annual Bayesian Modeling Applications Workshop (BMAW 2017)
will be a one-day workshop held in Sydney, Australia in conjunction with
the 33rd International Conference on Uncertainty in AI (UAI-2017
<http://auai.org/uai2017/index.php>).

The full Call for Papers is available at http://bmaw2017.azurewebsites.net/.

Key dates:
Submission: June 30th
Notification: July 10th
Full paper due: July 28th (Note: also ok to only submit abstract)
Workshop: August 15th.

Continuing a successful tradition as part of the UAI conference, the
Fourteenth Annual Bayesian Modeling Applications Workshop (BMAW 2017) will
provide a forum for exchange about real-world problems among applications
practitioners, tool developers, and researchers. The aim of the workshop is
to foster discussion on the challenges of building working applications of
probabilistic methods whilst considering stakeholders, user interaction,
tools, knowledge elicitation, learning, validation, system integration, and
deployment.

The practicality of sharing application development experience often comes
down to sharing reusable software: one is much more likely to add a brick
to the wall when she doesn’t have to re-make the other bricks. With models
more reliant on empirical setting of hyperparameters and training
schedules, reproducing the results can be impossible without access to
packaged original code, and even full detailed setups including minor
details of runtime environment. For example, deep learning models are often
released in full containers or virtual machines.

Code release is also a growing trend in our own community as seen by the
growing number of presentations on software in past workshops. The R
language CRAN repository just passed 10,000 packages earlier this year.
GitHub now contains millions of repositories. Search on most any algorithm,
and you are likely to find a repository where someone has at least made a
first attempt at implementing it.

BMAW 2017 is soliciting submissions describing real-world Bayesian
applications, with the desire that teams will explain what tools they
adopted to facilitate their work.

In addition to traditional full-length research papers, shorter abstracts
proposing a demonstration or tutorial are also welcome. Demonstrations are
appropriate to describe a probabilistic system or software addresing a
specific application. Tutorials should introduce novel probabilistic
software platforms or libraries enabling many applications.

We hope that teams will concurrently release links to the software
repositories and possibly data from their applications that can be reused.
Further, we solicit presentations on best practices and tools for sharing
data, code and models in easily reproducible ways, for example,
containerizing entire solutions.

While emphasizing the software and reproducibility aspects, we encourage
submissions from a broad spectrum of new and traditional application
topics, with a focus on probabilistic approaches, in distinction to the
discriminative machine learning and neural network methods that have gained
popularity elsewhere.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://abnms.org/pipermail/abnms_abnms.org/attachments/20170628/46692a77/attachment.html>


More information about the ABNMS mailing list