Artificial Societies and Computational Markets
(May 9)
Up to Date Program and Official Web Page: http://www.cs.toronto.edu/~grigoris/ascma.html
Description
Many natural and artificial systems in physical and social sciences consist
of entities (be they genes, cells, particles, neurons, humans or agents)
that are endowed with bounded rationality, in terms of information and
computational capacity, and potentially different operational goals. Based
on interaction mechanisms among their entities, these systems exhibit,
through time, a distributed organization that achieves goals at an aggregate
level such as survival, order, cognition and culture.
Despite the boundedness and locality of processing, these systems are
scalable to complex tasks and robust to uncertainties in the environments
within which they are situated. These contrasting and yet desirable problem-solving
properties have inspired metaphors, on the one hand for computer-based
modelling of systems in fields such as population genetics, statistical
mechanics, quantitative sociology and experimental economics; and on the
other, for the design and development of computational systems such as
cellular automata, computational markets and contract nets in Artificial
Life, Machine Learning and Distributed Artificial Intelligence.
The purpose of this workshop is to bring together researchers from
the above diverse fields in order to: (i) examine the design principles
and performance characteristics of various approaches on artificial societies
and computational markets and (ii) increase the cross-fertilization of
ideas on these computational models across domains. Submitted papers should
motivate the raison d'etre of such multiagent systems by referring to the
properties and complexity of the problem-solving task that is being addressed.
In particular, research questions relevant to the workshop include the
following:
-
how is bounded rationality and the interaction mechanism realized within
the multiagent system;
-
what levels of abstraction are used in the architecture of the multiagent
system;
-
what knowledge representation formalisms, languages and protocols are used
to develop the system;
-
what experimental procedures are appropriate for examining the emergent
behaviors and evaluating the performance of the system at the aggregate
as well as agent level;
-
what are the properties of the solutions provided by the system in terms
of local optima, stability and robustness;
-
what are the computational advantages gained by the proposed multiagent
model when compared with a single-agent model;
-
how can models such as artificial societies and computational markets provide
the framework for distributed software design and development.
Submission Details
The workshop will consist of invited talks, presentation and discussion
sessions and hands-on demonstrations. Researchers from artificial intelligence,
complex systems, quantitative sociology and experimental economics who
are interested in presenting their work should send a short paper (5-8
pages) describing work in progress or completed work. Persons desiring
to participate should submit position papers (up to two pages). We would
like to encourage submissions of video demonstrations, and working systems
that can be used for hands-on demonstration. Also of interest are review
papers (5-8 pages) that address theoretical and practical linkages among
frameworks of artificial societies and computational markets over different
fields. Papers should be e-mailed in PostScript form to karakoul@cibc.ca;
they must include: author's name(s), affiliation, complete mailing
address, phone number, fax number and email address.
Submissions for the workshop are due by January 15, 1998.
Notification of acceptance/rejection will be e-mailed by February 15, 1998.
Organizing Committee
Robert Axtell, Brookings Institute, USA
Richard Belew, University of California San
Diego, USA
Dave Cliff, MIT, USA
Innes Ferguson, Zuno Ltd., UK
Bernardo Huberman, Xerox Parc, USA
Grigoris Karakoulas (chair), CIBC & University
of Toronto, Canada
Blake LeBaron, University of Wisconsin, USA
John Miller, Carnegie Mellon University, USA
Chris Preist, HP Research Laboratories, HP,
UK
Tuomas Sandholm, Washington University, USA
Yoav Shoham, Stanford University, USA
Last Modified on