Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!hudson.lm.com!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!Germany.EU.net!EU.net!sun4nl!ruuinf!linda
From: linda@cs.ruu.nl (Linda van der Gaag)
Subject: CFP: IJCAI-95 Workshop on Building Probabilistic Networks
Message-ID: <D1u7xK.sF@cs.ruu.nl>
Sender: usenet@cs.ruu.nl (Six O'Clock News)
Organization: Utrecht University, Dept. of Computer Science
Date: Tue, 3 Jan 1995 16:17:43 GMT
Lines: 215



         PRELIMINARY ANNOUNCEMENT AND CALL FOR PAPERS
 
===========================================================================
           Fourteenth International Joint Conference
             on Artificial Intelligence (IJCAI-95)
                      Montreal, Canada
 
                        Workshop on
Building Probabilistic Networks: Where Do the Numbers Come From?
                   Monday, August 21, 1995
 
             Submissions due on March 15, 1995
---------------------------------------------------------------------------
 
 
WORKSHOP THEME:
 
Probabilistic networks are now well established as effective and practical
representations of knowledge for reasoning under uncertainty, as the
increasing number of successful applications in such domains as diagnosis,
planning, learning, vision, and natural language processing demonstrates.
A probabilistic network (also referred to as belief network, Bayesian
network, or causal network) consists of a graphical part, encoding a
domain's variables and the qualitative dependence relations among them,
and a quantitative part, encoding a probability distribution over these
variables.  The task of eliciting the graphical part from domain experts
is comparable to knowledge engineering for other AI representations and,
although it may require significant effort, is generally considered quite
doable.  The task of obtaining the probabilities often appears more
daunting to those embarking on constructing a probabilistic network:
"Where do the numbers come from?" is a commonly asked question.  This
question is the central theme of the workshop.
 
For building probabilistic networks for data-rich application domains,
methods are available for estimating the probabilities for the network
automatically from data.  When there is little or no data available, one
might resort to standard methods developed and used by decision analysts
to elicit judgmental probabilities from domain experts.  However,
application of these methods to the construction of probabilistic
networks is hampered by several obstacles.  The foremost is the large
size and complex dependence structure of probabilistic networks, relative
to most conventional decision analytic models.  A subtle, and often
controversial, issue is the need to encode causality, which is often
deemed irrelevant in decision analysis.  The need for real-time reasoning
further requires simplifying assumptions for computational tractability.
In fact, effective use of probabilistic networks requires careful tradeoff
between the desire for a large and rich model to obtain accurate
performance and the costs of construction, inference, and maintenance of
large models.  Researchers and practitioners in the use of probabilistic
networks have developed effective and practical techniques to handle
these problems.  Few of these techniques, however, have become widely
available general purpose tools and little is known as to their
compatibility.
 
 
WORKSHOP OBJECTIVE:
 
In this workshop, we would like to provide a forum for a stimulating
exchange of information among a group of researchers with experience
in building probabilistic networks.  The main objective of the workshop
is to identify methods of eliciting probabilistic information that are
general enough to be shared, combined, and developed further.  In
addition, we hope that the workshop will lead to formulation of an
agenda for making these methods available to the AI community.
 
 
WORKSHOP ISSUES:
 
We invite papers from researchers with experience in building probabilistic
networks.  We are seeking information about how they obtained the structure
and the numbers in their networks, what problems they encountered, whether
and how they solved them, what worked, what did not, and why.
 
Examples of specific technical issues that the workshop will focus on are:
various sources of probabilistic information; granularity of the model
versus granularity of the domain knowledge; good and bad experiences with
collecting data; identification of domain characteristics that allow for
applying specific elicitation methods; elicitation of qualitative knowledge
and combining it with quantitative information; the effects of rough versus
fine-tuned probability assessments on performance; the sensitivity of
probabilistic reasoning to the precision of the probabilities; bringing
experiences with existing AI knowledge elicitation techniques to building
probabilistic networks; computer aids for probability assessment; treatment
of distributions that vary over time; dealing with vagueness, imprecision,
inconsistency, and ignorance when building probabilistic networks.
 
We strongly encourage submission of papers that discuss practical
experiences, although we are open for important theoretical contributions
that are directly applicable in practice.
 
 
SUBMISSION GUIDELINES:
 
Papers should be limited to a total of 15 pages including a title page,
tables, figures, and references.  Papers should be printed on 8.5" x 11"
or A4 sized paper using 12 point type (10 characters per inch) with a
one inch margin and no more than 43 lines per page.  The title page must
include the names, full postal and e-mail addresses, phone and FAX
numbers of all authors, an abstract (not more than 200 words), and a
list of keywords reflecting the topics addressed by the paper.
 
Electronic submissions are strongly encouraged; the preferred formats
are self-contained (!) LaTeX and PostScript.  If electronic submission
is absolutely impossible, please send four clearly legible hard copies
of papers to the primary contact mentioned below; FAX submissions are
not acceptable.
 
Primary contact:
Linda C. van der Gaag
  Utrecht University
  Department of Computer Science
  P.O. Box 80.089
  3508 TB Utrecht, The Netherlands
  e-mail: linda@cs.ruu.nl
  Phone: +31-30-534089
  FAX:   +31-30-513791
 
All communication about submissions will be electronic: confirmation
of reception of papers as well as notification of acceptance or
rejection will be sent by e-mail.
 
All accepted papers will be published in the workshop's working notes
and will be distributed in hard copy by IJCAI at the workshop.  We will
strive to make the accepted papers available electronically late June
1995.  Upon sufficient interest, we intend to publish revised versions
of a selection of papers in book form.  The authors should take this
into account while preparing their papers.
 
Upon confirmation of participation, each attendee of the workshop will be
expected to provide electronically a one-page vita including research
interests, ongoing projects, contact addresses, and a short list of
most relevant publications.  We intend to distribute these vitae among
all participants to facilitate interaction.
 
 
SELECTION CRITERIA:
 
Each submitted paper will be reviewed by at least two reviewers and
will be judged on significance, originality, and clarity.  In selecting
the papers, we will also consider breadth of coverage of the main theme
of the workshop.
 
Participation in the workshop will be by invitation.  Authors of accepted
papers will be automatically invited to attend the workshop.  We also
expect to be able to accommodate a small number of additional researchers
whose presence and experience in building probabilistic models can
contribute significantly to the discussions in the workshop.  Researchers
who are interested in participating in the workshop without presenting
a paper are requested to submit a one-page vita as outlined above to
the primary contact.
 
 
IMPORTANT DATES:
 
Submission deadline         : March 15, 1995
Notification of acceptance  : April 15, 1995
Camera-ready copy due and
confirmation of attendance  : April 28, 1995
Workshop date               : August 21, 1995
 
 
WORKSHOP FORMAT:
 
The workshop will be one day long and will take place on Monday, August 21,
between the Eleventh Conference on Uncertainty in Artificial Intelligence
and the IJCAI-95 main conference.  The workshop will be divided into three
or four sessions.  There will be a small number of carefully selected
presentations and a panel discussion at the conclusion of the workshop.
We plan to have an informal dinner get-together in one of the neighboring
restaurants.
 
As mentioned before, participation in the workshop will be by invitation
only.  The number of participants will be limited to 30.  All workshop
participants will be expected to register for the main IJCAI-95 conference.  In addition, there will be a separate workshop registration fee of $50.
 
 
ORGANIZING COMMITTEE:
 
If you have any questions about the workshop, please contact
one of the members of the organizing committee:
 
Marek J. Druzdzel (co-chair)
  University of Pittsburgh
  Department of Information Science
  135 North Bellefield Avenue
  Pittsburgh, PA 15260, U.S.A.
  e-mail: marek@lis.pitt.edu
  Phone: +1-412-624-9432
  FAX  : +1-412-624-2788
 
Linda C. van der Gaag (co-chair)
  Utrecht University
  Department of Computer Science
  P.O. Box 80.089
  3508 TB Utrecht, The Netherlands
  e-mail: linda@cs.ruu.nl
  Phone: +31-30-534089
  FAX:   +31-30-513791
 
Max Henrion
  Lumina Decision Systems, Inc.
  4984 El Camino Real, Suite 105
  Los Altos, CA 94022, U.S.A.
  e-mail: henrion@lumina.com
  Phone: +1-415-254-0189
 
Finn V. Jensen
  Aalborg University
  Department of Mathematics and Computer Science
  Fredrik Bajers Vej 7-E
  DK-9220 Aalborg OE, Denmark
  e-mail: fvj@iesd.auc.dk
  Phone: +45-98-158522
