Title: Adaptive Load Balancing: a study in co-learning

Authors: Andrea Schaerf, Yoav Shoham and Moshe Tennenholtz
         aschaerf@dis.uniroma1.it, shoham@flamingo.stanford.edu,
         moshet@ie.technion.ac.il

Abstract: 

This is a short abstract submitted to the Adaptation and Learning
in Multiagent Systems workshop.  A much longer and detailed
version of this paper, called "Adaptive Load Balancing: A Study
in Multi-Agent Learning", will appear in the Journal of
Artificial Intelligence Research.

In our work we study the process of multi-agent reinforcement
learning in the context of load balancing in a distributed system
without use of either central coordination or explicit
communication.  We first define a precise framework in which to
study adaptive load balancing, improtan features of which are its
stochastic nature and the purely local inforamtion available to
individual agents.  Given this framework, we show illuminating
results on the interplay between basic adaptive beehavior
parameters and their effect on system efficiency.  We then
investigate the properties of adaptive load balancing in
heterogeneous populations, and expose various phenomena related
to the issue of exploration vs. exploitation in that context.  In
addition, we show that naive use of communication may not
improve, and might even deteriorate system efficiency.