Evolving cases for case-based reasoning multiagent negotiations
Leen‐Kiat Soh, Costas Tsatsoulis, Mara Elizabeth Jones, Arvin Agah · 2001
The work reported in this paper applies genetic algorithms (GAs) to the automatic generation of cases for a case-based reasoning (CBR) system employed for multiagent negotiations. Our problem domain is in resource allocation and constraint satisfaction. In particular, our multiagent system has a set of agents working cooperatively to track multiple moving targets as accurately as possible. Each agent maintains its local information base and has a limited sensor capability (each sensor can only cover a small area of the environment and accurate target tracking requires at least three sensors). Thus, the agents controlling the sensors have to be communicative—to inform their neighbors of incoming targets and to ask them to perform certain tasks. Moreover, the agents may share the same CPU platform and thus need to re-allocate their usage of the limited CPU resources according to their tasks and their perceived environments. This also motivates the agents to share local constraints and cooperate.