|
(See ISLA for the new IAS website. Most maintenance activities on this old website have stopped.)
Multiagent coordination
Objective
Within a group of cooperating agents the decision making
of an individual agent depends on the actions of the other
agents. Consequently, the joint state-action space of the agents
scales exponentially with the number of agents, making traditional
decision making models intractable. The framework of coordination
graphs
(Guestrin, Koller, and Parr, 2002) allows for a tractable approach
to multiagent coordination, by decomposing the global payoff function
of the system into a sum of local terms. Each term depends on few
agents only, which allows for efficient coordination.
Our work on coordination graphs involves:
-
Message passing techniques for approximate multiagent
decision making (similar to belief propagation in Bayesian networks).
-
Distributed cooperative reinforcement learning (Q-learning) using
coordination graphs.
Research group members
drs. J.R. Kok
dr. N. Vlassis
Funding
This research is partially supported by PROGRESS, the
embedded systems research program of the Dutch organization for Scientific
Research NWO, the Dutch Ministry of Economic Affairs and the Technology
Foundation STW, project AES.5414.
More information
More information (references, etc.) can be found here.
|
|
