Effects of Local Information on Group Behavior

Sandip Sen, Shounak Roychowdhury and Neeraj Arora, University of Tulsa, USA

The performance of individual agents in a group depends critically on the quality of information available to it about local and global goals and resources. In general it is assumed that the more accurate and up-to-date the available information, the better is the expected performance of the individual and the group. This conclusion can be challenged in a number of scenarios. We investigate the use of limited information by agents in choosing between one of several different options, and conclude that if agents are deliberately kept ignorant about any number of options, the entire group can converge faster to a stable and optimal configuration. We also demonstrate how a couple of coalition formation schemes improves the rate of convergence and conclude that a variable, rather than fixed, coalition formation mechanism is more effective.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.