Prof. Raja Speaks at IIT Madras
January 09, 2015
Individual agent policies can sometimes lead to inconsistent agent actions in a multiagent context. These are especially problematic where inconsistencies could be mission-critical in that they violate hard domain constraints like trying to use the same resource for two different activities. It is important to coordinate the policies in a cost effective way without over-constraining individual agents. Professor Raja describes some of our ongoing research that harnesses game theory, decentralized learning and topological information of agent networks for coordinating policies. Professor Raja also discusses the effectiveness of our algorithms in the context of a weather tracking domain and language coordination application.