Adaptive choice of behavior and protocol parameters

Authors:
Frank Grove;Sandip Sen;Oly Mistry
Affiliations:
University of Tulsa, Tulsa, OK;University of Tulsa, Tulsa, OK;University of Tulsa, Tulsa, OK
Venue:
PRIMA'10 Proceedings of the 13th international conference on Principles and Practice of Multi-Agent Systems
Year:
2010

Citing 5
Cited 0

Agents that buy and sell

Communications of the ACM
Strategic negotiation in multiagent environments

Strategic negotiation in multiagent environments
Socially Conscious Decision-Making

Autonomous Agents and Multi-Agent Systems
Bargaining with incomplete information

Annals of Mathematics and Artificial Intelligence
Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations

Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations

Quantified Score

Hi-index	0.00

Visualization

Abstract

Research on interaction between multiple self-interested agents has focused on either designing rational behavior for agents given the interaction protocol or designing the interaction protocol that will promote desirable rational behavior by agents. We believe that in certain situations self-interested agents can be interested in both choosing desirable protocols and deciding effective strategies to follow under the chosen protocol. We experiment with a market situation where agents repeatedly negotiate to decide on the allocation of indivisible resources. We present a parameterized protocol selection scheme which can be used by agents to select the interaction protocol to use. We show that learning agents can greatly improve performance by adapting the protocol used and the behavior adopted against a range of opponents.