computer-go: AIPS-02 Workshop on Planning and Scheduling with Multiple Criteria

[Martin Mueller <mmueller@xxxxxxxxxxxxxx>]

Hello,
this announcement may only be of interest for a few of you. It is about
planning with multiple criteria, which sounds like a useful idea in a
complex game such as Go.

	Martin

                AIPS-02 Workshop on
Planning and Scheduling with Multiple Criteria
        Toulouse, France, April 23-27 2002
         http://www.csd.auth.gr/~lpis/aips02

                Sponsored by PLANET

Call for Papers/Participation

Most real-world problems demand the consideration of many criteria, such as
plan duration, resource consumption, profit, safety etc, either separately,
or in some combination. In the former case the plans are optimized for a
single criterion and the other criteria are handled as constraints, whereas
in the latter case the plans are optimized with respect to an arbitrary
combination of the criteria. In many cases the criteria are in conflict and
a trade off must be identified. For example, in a manufacturing domain the
criteria may be to maximize the work in progress (to maximize the number of
orders fulfilled) and minimize inventory (to minimize the amount of raw
materials purchased) but to fulfill a large number of orders a large
inventory must be kept. In addition to resolving conflicts several issues
arise when taking into account multiple criteria, such as defining
optimality, expressing preferences, aggregating the criteria, generating
bounds and/or heuristic distance information, guiding search, pruning
branches, trading off planning time and solution optimality, etc.
Dealing with multiple criteria is not a unique problem faced by researchers
in AI planning and scheduling. Evaluating states and solutions based on
multiple criteria is a problem occurring in other fields, in particular,
combinatorial game search and multi-criteria decision making. Researchers
in these areas have tended to address these related problems from a search
or operations research perspective, respectively.
During the last few years significant improvements have been made in the
capabilities of planning systems to the point that they are now capable of
producing plans with hundreds of actions in a few seconds. While such
performance is commendable, it has been achieved with very simple action
descriptions that would have little applicability on real-world problems.
We believe that it is the time to investigate ways of improving action
descriptions and to handle reasoning with multiple criteria, an area that
has been neglected for too long.

The workshop has several goals:
1. to review the current state of the art in reasoning with multiple criteria
2. to initiate discussions within the AI planning and scheduling
communities on how these problems may be addressed
3. to initiate the transfer of applicable techniques, insights and
experiences from other communities such as Operations Research, Uncertainty
and Game communities.

Technical Part
==============
The technical part of the workshop will consist of papers dealing with
multiple criteria in planning, scheduling, constraint reasoning, decision
making, uncertainty and game search and will focus on, but will not be
limited to addressing the following research questions:
* defining optimality in the presence of multiple criteria
* representing criteria preferences
* computing aggregate criteria
* generic methods to define bounds for multi-criteria branch-and-bound search
* techniques to guide search based on multiple criteria
* computing heuristic distance information in the presence of more than one
criterion
* near-optimal solutions, hard and soft criteria
* evaluation of planners' performance based on multiple criteria (e.g. in
the planning competition)
* experience in dealing with real world applications

Invited Talks & Panels
======================
To provide a focus for the interactions and discussions we plan to have two
invited speakers, one from the planning community having experience in
planning with multiple criteria, and one from the game search community.
These talks will summarize the state of the art in their respective fields
with respect to the issue of dealing with multiple criteria and highlight
promising research paths.

Submissions
Papers should be a maximum of 5000 words, and should be submitted by email
to Ioannis Refanidis (yrefanid@xxxxxxxxxxxxxxxxx) in any convenient format (PDF,
PostScript, MS-Word), preferred compressed with gzip or winzip.

Critical Dates
==============
* Paper Submission Deadline: February 11, 2002
* Paper Notification: March 15, 2002
* Final papers Due: March 22, 2002

Organizers
==========
Brian Drabble, CIRL, (drabble@xxxxxxxxxxxxxxxxx)
Jana Koehler, IBM Zurich, (koe@xxxxxxxxxxxxxxxxx)
Ioannis Refanidis, Aristotle Unversity (yrefanid@xxxxxxxxxxxxxxxxx)

Committee
=========
Mark Boddy, Honeywell
Yannis Dimopoulos, University of Cyprus
Patrick Doherty, Linkoping University
Alfonso Gerevini, University of Brescia
Hector Geffner, Universidad Simon Bolivar
Richard Goodwin, IBM T.J.Watson Center
Martin Mueller, University of Alberta
Nicola Muscettola, NASA
Karen Myers, SRI International
Alexis Tsoukias, LAMSADE, Universite Paris Dauphine
Ioannis Vlahavas, Aristotle University
Joachim Paul Walser, I2 Technologies