%0 Conference Paper %1 conf/aamas06/Diuk %A Diuk, Carlos %A Strehl, Alexander L. %A Littman, Michael L. %B AAMAS '06: Proceedings of the fifth international joint conference on Autonomous agents and multiagent systems %C New York, NY, USA %D 2006 %I ACM Press %K aamas06 agents hierarchical planning %P 313--319 %R http://doi.acm.org/10.1145/1160633.1160686 %T A hierarchical approach to efficient reinforcement learning in deterministic domains %U http://portal.acm.org/citation.cfm?id=1160633.1160686&coll=ACM&dl=ACM&type=series&idx=1160633&part=Proceedings&WantType=Proceedings&title=International%20Conference%20on%20Autonomous%20Agents&CFID=25754815&CFTOKEN=92537800 %X Factored representations, model-based learning, and hierarchies are well-studied techniques for improving the learning efficiency of reinforcement-learning algorithms in large-scale state spaces. We bring these three ideas together in a new algorithm. Our algorithm tackles two open problems from the reinforcement-learning literature, and provides a solution to those problems in deterministic domains. First, it shows how models can improve learning speed in the hierarchy-based MaxQ framework without disrupting opportunities for state abstraction. Second, we show how hierarchies can augment existing factored exploration algorithms to achieve not only low sample complexity for learning, but provably efficient planning as well. We illustrate the resulting performance gains in example domains. We prove polynomial bounds on the computational effort needed to attain near optimal performance within the hierarchy. %@ 1-59593-303-4

@inproceedings{conf/aamas06/Diuk, abstract = {Factored representations, model-based learning, and hierarchies are well-studied techniques for improving the learning efficiency of reinforcement-learning algorithms in large-scale state spaces. We bring these three ideas together in a new algorithm. Our algorithm tackles two open problems from the reinforcement-learning literature, and provides a solution to those problems in deterministic domains. First, it shows how models can improve learning speed in the hierarchy-based MaxQ framework without disrupting opportunities for state abstraction. Second, we show how hierarchies can augment existing factored exploration algorithms to achieve not only low sample complexity for learning, but provably efficient planning as well. We illustrate the resulting performance gains in example domains. We prove polynomial bounds on the computational effort needed to attain near optimal performance within the hierarchy.}, added-at = {2007-08-03T14:48:24.000+0200}, address = {New York, NY, USA}, author = {Diuk, Carlos and Strehl, Alexander L. and Littman, Michael L.}, biburl = {https://www.bibsonomy.org/bibtex/2559c17efdf6d65b594fa1ed92d170f7c/mpfingst}, booktitle = {AAMAS '06: Proceedings of the fifth international joint conference on Autonomous agents and multiagent systems}, description = {: AAMAS '06, A hierarchical approach to ...}, doi = {http://doi.acm.org/10.1145/1160633.1160686}, interhash = {3cb6771ee6cde726aa9f0bca629f7e3d}, intrahash = {559c17efdf6d65b594fa1ed92d170f7c}, isbn = {1-59593-303-4}, keywords = {aamas06 agents hierarchical planning}, location = {Hakodate, Japan}, pages = {313--319}, publisher = {ACM Press}, timestamp = {2007-08-03T14:48:24.000+0200}, title = {A hierarchical approach to efficient reinforcement learning in deterministic domains}, url = {http://portal.acm.org/citation.cfm?id=1160633.1160686&coll=ACM&dl=ACM&type=series&idx=1160633&part=Proceedings&WantType=Proceedings&title=International%20Conference%20on%20Autonomous%20Agents&CFID=25754815&CFTOKEN=92537800}, year = 2006 }