An intrinsic reward mechanism for efficient exploration Export

@inproceedings{simsek-barto:icml2006, abstract = {How should a reinforcement learning agent act if its sole purpose is to efficiently learn an optimal policy for later use? In other words, how should it explore, to be able to exploit later? We formulate this problem as a Markov Decision Process by explicitly modeling the internal state of the agent and propose a principled heuristic for its solution. We present experimental results in a number of domains, also exploring the algorithm's use for learning a policy for a skill given its reward function---an important but neglected component of skill discovery.}, address = {New York, NY, USA}, author = {\c{s}im\c{s}ek, \"{O}zg\"{u}r and Barto, Andrew G.}, booktitle = {ICML '06: Proceedings of the 23rd international conference on Machine learning}, citeulike-article-id = {2604098}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1143949}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1143844.1143949}, doi = {10.1145/1143844.1143949}, isbn = {1595933832}, keywords = {icml06, reinforcement\_learning}, pages = {833--840}, posted-at = {2008-03-27 22:33:44}, priority = {2}, publisher = {ACM}, title = {An intrinsic reward mechanism for efficient exploration}, url = {http://dx.doi.org/10.1145/1143844.1143949}, year = {2006} }

TY - CONF ID - simsek-barto:icml2006 L3 - citeulike-article-id:2604098 N2 - How should a reinforcement learning agent act if its sole purpose is to efficiently learn an optimal policy for later use? In other words, how should it explore, to be able to exploit later? We formulate this problem as a Markov Decision Process by explicitly modeling the internal state of the agent and propose a principled heuristic for its solution. We present experimental results in a number of domains, also exploring the algorithm's use for learning a policy for a skill given its reward function---an important but neglected component of skill discovery. SN - 1595933832 AD - New York, NY, USA EP - 840 TI - An intrinsic reward mechanism for efficient exploration PB - ACM T2 - ICML '06: Proceedings of the 23rd international conference on Machine learning SP - 833 KW - icml06 KW - reinforcement_learning AU - \c{s}im\c{s}ek, Özgür AU - Barto, Andrew PY - 2006/// UR - http://dx.doi.org/10.1145/1143844.1143949 ER -