Multiagent Reinforcement Learning Algorithm by Dynamically Merging Markov Decision Processes
Full Text: ghavamzadeh02multiagent.pdfOne general strategy for accelerating the learning of cooperative multiagent tasks is to reuse (good or optimal) solutions to the task when each agent is acting alone. In this paper, we formalize this approach as dynamically merging solutions to multiple Markov decision processes (MDPs), each representing an individual agent's solution when acting alone, to obtain (good or optimal) solutions to the overall multiagent MDP when all the agents act together. We present a new temporal-difference.
Citation
M. Ghavamzadeh, S. Mahadevan. "Multiagent Reinforcement Learning Algorithm by Dynamically Merging Markov Decision Processes". Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), pp 845-846, July 2002.Keywords: | |
Category: | In Conference |
BibTeX
@incollection{Ghavamzadeh+Mahadevan:AAMAS02, author = {Mohammad Ghavamzadeh and Sridhar Mahadevan}, title = {Multiagent Reinforcement Learning Algorithm by Dynamically Merging Markov Decision Processes}, Pages = {845-846}, booktitle = {Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS)}, year = 2002, }Last Updated: June 11, 2007
Submitted by Staurt H. Johnson