Multiagent systems and complex systems simulation: Why and how?

Pierre Marcenac
University of La Reunion, France

Friday, 24th April 1998
1130 - 1230

Room 10.9.33
RMIT City campus
Swanston Street, Melbourne.

Abstract:

Agents and multiagent systems are becoming an important area of research and development in Artificial Intelligence today.  Agents are typically knowledge representation entities characterised by independence and autonomy.  When aggregated in systems (called multiagent systems), they provide a powerful computational tool, for which dynamic aspects are based on interactions with others, rather than centralised control.  Agents are particularly adapted to complex systems modelling, in which environments are unpredictable and simulation required.  The essential contribution of the agent paradigm in this context is indisputably the concept of emergence.

After reviewing some related definitions of the agent, the seminar will discuss the benefits of the agent paradigm in such a context, through emergence.  An example of an application developed in our team will be used to illustrate it.  Even if the example has been applied in the volcano eruption prediction area, it could easily be translated in different areas.

Biosketch:

Dr. Pierre Marcenac is an Associate Professor at IREMIA, University of La Reunion, France.  He completed a PhD in Intelligent Tutoring Systems at the University of Nice-Sophia Antipolis (France) in 1990.  His research interests include knowledge representation and multiagent systems.