TY - GEN
T1 - Adaptive learning to optimize resource management in a multi-agent framework
AU - Chen, Chao
AU - Soh, Leen Kiat
PY - 2004
Y1 - 2004
N2 - In a collaborative multi-agent framework where the agents provide services to their local customers, they are encouraged to work collaboratively to improve group services as a community. The dual objectives raise the question of how agents should distribute their local resource given large number of tasks from the local customer and the community. In this paper, we propose an adaptive learning mechanism that can be embedded into an agent's decision making as an individual as well as a member of a team. It enables agents to make decisions based on the observation of the current quality of service and the distribution of the different types of tasks encountered in the past. In turn, an agent estimates the future task distribution and decides how to handle tasks initiated in the community. From the task initiator's point of view, the learning is to observe the service quality and reduce redundant initiations. From the task responder's point of view, the goal is to distribute the amount of its resources to the most needed tasks.
AB - In a collaborative multi-agent framework where the agents provide services to their local customers, they are encouraged to work collaboratively to improve group services as a community. The dual objectives raise the question of how agents should distribute their local resource given large number of tasks from the local customer and the community. In this paper, we propose an adaptive learning mechanism that can be embedded into an agent's decision making as an individual as well as a member of a team. It enables agents to make decisions based on the observation of the current quality of service and the distribution of the different types of tasks encountered in the past. In turn, an agent estimates the future task distribution and decides how to handle tasks initiated in the community. From the task initiator's point of view, the learning is to observe the service quality and reduce redundant initiations. From the task responder's point of view, the goal is to distribute the amount of its resources to the most needed tasks.
UR - http://www.scopus.com/inward/record.url?scp=12744281513&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=12744281513&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:12744281513
SN - 1932415335
SN - 9781932415339
T3 - Proceedings of the International Conference on Artificial Intelligence, IC-AI'04
SP - 386
EP - 389
BT - Proceedings of the International Conference on Artificial Intelligence, IC-AI'04
A2 - Arabnia, H.R.
T2 - Proceedings of the International Conference on Artificial Intelligence, IC-AI'04
Y2 - 21 June 2004 through 24 June 2004
ER -