Applications of multi-agent learning in e-commerce and
			 autonomic computing

Jeff Kephart
IBM Research

During the next decade and beyond, electronic commerce and autonomic
(self-managing) computing are likely to be two of the most broad and
important realms of application for multi-agent learning.  In this
talk, I will survey a variety of fundamental issues that arise in each
of these contexts.

We foresee the world economy as a competitive multi-agent system of
unprecedented scale, with billions of economically-motivated software
agents representing individuals and corporations purchasing and
pricing goods and services alongside their human counterparts.  To
provide a glimpse of this future, I will describe some experiments
involving humans and software agents that use various adaptive pricing
and bidding strategies, and discuss an approach to understanding
multi-agent strategic interactions in complex economic games.

We envision autonomic computing systems as large-scale cooperative
multi-agent systems that are characterized by self-configuration,
self-healing, and self-optimization.  I will describe several major
research challenges in multi-agent learning that must be met if we are
to realize the vision of autonomic computing, including automated
negotiation and conflict resolution.