Track:
Multiagent Systems
Downloads:
Abstract:
In order to effectively exploit opportunities presented in the environment agents in a group must be well-adapted to each other and to the environment. Agents that fail to adapt and modify their behavior to suit environmental demands can hinder, rather than aid, in achieving group goals. Adaptability and flexibility are key components of intelligent behavior which allow agent groups to improve performance in a given domain using prior problem solving experience. This paper focuses on a particular incremental learning mechanism by which agents can better adapt to each other using problem solving experience. In particular, we propose a framework in which individual group members learn cases to improve their model of other group members. Using these models agents can choose less greedy and more appropriate actions in the context of the group. We use a testbed problem from the distributed AI literature to show that simultaneous learning by group members can lead to significant improvement in group performance and efficiency over agent groups following static behavioral rules.