At the end of the course, a student should be able to design and build
a multi-agent system as a third semester project.
Conventional AI systems such, as expert systems, are closed systems in that
they make decisions with minimal external input. These systems fail
miserably when presented with problems outside their limited field
of expertise. The traditional answer from AI is to propose new forms
of knowledge representation and/or to accumulate vast quantities of
common sense knowledge in one system. The new frontier
Multi-agent Systems (MAS), proposes an open systems approach
by building societies of agents (herds of robots) for which
the whole is more than the sum of the parts.
Whereas conventional AI draws its inspiration from neurophysiology, psychology
and mathematical logic, MAS has sociology, anthropology, economics, operations
research, control theory, systems science and management science as additional metaphors.
The explosion of interest in MAS has followed the explosion of
interest in the Internet and WWW. This interest is witnessed by
the July 1994 issue of Comm ACM. The year, 1995, saw the
First International Conference on Multi-Agent Systems.
The course is divided into two parts, micro and macro-theories.
The first part of the course focuses on micro-systems: the
architecture of an individual agent and how it makes decisions.
This part of the course will draw heavily on the essential text.
This text provides good support for those with little background
The second part of the course is about macro-systems, where
the concern is interagent dynamics. This part of the course will
work from the research papers cited below.
A three hour exam in April will contribute 70% to the total for the
course. Coursework, contributing a total of 30%, will consist of
two group projects, one for each half of the course.
- Malone TW and Crowston K (1996) The interdisciplinary study of coordination,
ACM Computing Surveys 26M(1)87-119
- Scalabrin EE et al (1996) A generic model of cognitive agents to develop open
system, Springer LNCS 1159:61-70
- Lesser VR (1990) An overview of DAI: viewing distributed AI as distributed search,
Journal of Japense Society for AI 5(4)
- Davis R and Smith RG (1983) Negotiation as a metaphor for distributed problem solving,
Artificial Intelligence 20:63-109
- Zlotkin G and Rosenschein JS (1989) Negotiation and task sharing among autonomous
agents in cooperative domains, 11th IJCAI, 912-17
- Zlotkin G and Rosenschein JS (1991) Incomplete information and deception in
multi-agent negotiation, 12th IJCAI, 225-31
- Ephrati E and Rosenschein JS (1991) The Clarke Tax as a consensus mechanism among
automated agents, AAAI-91, 173-78
- Fox MS (1988) An organizational view of distributed systems, IEEE Trans on
Systems, Man and Cybernetics 11(1)70-80
- Miller MS and Drexler EK (1988) Markets and computation: agoric open systems,
in Huberman BA ed, The Ecology of Computation, Elsevier, 133-76
- Waldspurger CA et al (1992) Spawn: a distributed computational economy, IEEE Trans
on Software Engineering 18(2)103-16
- Lieberman H (1987) Concurrent object-oriented programming in Act1, in Yonezawa A and Tokoro M eds, Object Oriented Concurrent Programming, MIT Press, 9-36. (1987)
- Hewitt C and Inman J (1991) DAI betwixt and between: from intelligent agents to
open systems science, IEEE Trans on Systems, Man and Cybernetics 21(6)1409-19
- Kahn KM and Miller MS (1988) Language design and open systems, in Huberman BA,
The Ecology of Computation, Elsevier, 291-313
- Shoham Y (1993) Agent-oriented programming, Artificial Intelligence 60:51-92
- Shoham Y and Tennenholtz M (1995) On social laws for artificial agent societies:
off-line design, Artificial Intelligence 73:231-52
- Genesereth MR and Ketchpel SP (1994) Software agents, CACM 37(7)48-53
More detailed information can be found on the courseware page.
All the DCS Information Sheets are available on our WWW pages.
Last modified by
gar on Thu Aug 20 1998