First International Full-day Workshop onMetareasoning in Agent-Based Systems
Monday, May 14, 2007
|
|
|
Metareasoning is the
process of reasoning about reasoning itself. It is composed of both the
meta-level control of computational activities and the introspective
monitoring of reasoning to evaluate and to explain computation. Meta-level
control is the ability of an agent to efficiently trade off its resources
between object level actions (computations) and ground level actions to
maximize the quality of its decisions. While meta-level control allows
agents to dynamically adapt their object level computation, it could
interfere with ground level performance. Identifying the decision points
that require meta-level control is of importance to the performance of
agents operating in resource-bounded environments. Introspective monitoring
is necessary to gather sufficient information with which to make effective
meta-level control decisions. Monitoring may involve the gathering of
computational performance data so as to build a profile of various decision
algorithms. When reasoning fails at some task, it may involve the
explanation of the causal contributions of failure and the diagnosis of the
reasoning process.
This workshop will explore various aspects of
metareasoning and its role in single-agent and
multiagent applications. There are significant research questions
about the extent to which meta-level control and monitoring affects
multiagent activity. In
multiagent systems, where the quality of joint decisions matter, the
value obtained by an agent exploring some portion of its decision space can
be dependent upon the degree to which other agents are exploring
complementary parts of their spaces. The problem of coordinated meta-level
control refers to this question of how agents should coordinate their
strategies to maximize the value of their joint actions.
The significance of this workshop is to bring together researchers from
different areas such as multi-agent systems, planning/scheduling, case-based
reasoning and cognitive science to gain broad insights into specific
research issues related to metareasoning in
agent-based systems. The goals are to foster discussions about ongoing
research, to establish directions for future research and collaborations,
and to identify best practices for evaluation of
metareasoning.
Relevant topics include, but are not limited to, the following:
• Meta-level control in single-agent systems
* Distributed meta-level control in multiagent
systems
* Meta-explanation and self-explanation
* Centralized versus distributed meta-level control
* Coordinated meta-level control
* The role of state abstraction in metareasoning
* Domains and/or problems where meta-level control is beneficial
* Evaluation of metareasoning systems
* Logical introspection and agents
* The role of metareasoning in building safe
and/or robust systems
* The integration of meta-level control and monitoring
* Learning agents and metareasoning
IMPORTANT DATES
|
Submission Deadline |
February 5, 2007 |
|
Acceptance Notification |
March 5, 2007 |
|
Camera-ready Copy |
March 19, 2007 |
|
Workshop |
May 14, 2007 |
SUBMISSION INSTRUCTIONS
We encourage the submission of high quality, original papers that are not submitted for publication elsewhere. The submission should not exceed 15 pages in the Springer-Verlag LNCS style (http://www.springer.de/comp/lncs/authors.html), either in PostScript or PDF format. Surface mail address, e-mail addresses should be included for all contributing authors. Short extended abstracts or position papers are also welcome. Submissions must be emailed to either chair (anraja@uncc.edu or mcox@bbn.com) by the deadline period.
We also intend to select
key papers from the workshop to be published in an edited collection.
Anita Raja,
Department of Software and Information Systems,
310 D Woodward Hall,
University of North Carolina at Charlotte, Charlotte, NC 28223.
Email: anraja@uncc.edu
Tel: (704) 687-8651
Fax: (704) 687-4893
Michael T. Cox,
Intelligent Computing,
BBN Technologies,
Cambridge, MA 02138
Email: mcox@bbn.com
Tel: (617)873-3632
Mike Anderson, Franklin & Marshall
College.
Mark Boddy,
Adventium Labs.
Maria Gini, University of
Minnesota.
Ashok Goel, Georgia
Institute of Technology.
Eric Hansen, Mississippi State University.
Kate Larson, University of Waterloo.
David Leake, Indiana
University.
Victor Lesser, University of Massachusetts.
Rajiv Maheswaran, University
of Southern California.
Abdel-illah
Mouaddib, University of Caen.
David Musliner, Honeywell Laboratories.
Martijn Schut,
Vrije Universiteit.
Stuart Shapiro, University at Buffalo.
Stephen Smith, Carnegie Mellon University
Tom Wagner, DARPA.
Shlomo
Zilberstein, University of Massachusetts.