A Bayesian Framework for Reinforcement Learning

@inproceedings{citeulike:229813, abstract = {The reinforcement learning problem can be decomposed into two parallel types of inference: (i) estimating the parameters of a model for the underlying process; (ii) determining behavior which maximizes return under the estimated model. Following Dearden, Friedman and Andre (1999), it is proposed that the learning process estimates online the full posterior distribution over models. To determine behavior, a hypothesis is sampled from this distribution and the greedy policy with respect ...}, author = {Strens, Malcolm }, booktitle = {Proc. 17th International Conf. on Machine Learning}, citeulike-article-id = {229813}, keywords = {bayesian, learning, reinforcement}, pages = {943--950}, posted-at = {2007-11-16 11:32:00}, priority = {2}, publisher = {Morgan Kaufmann, San Francisco, CA}, title = {A Bayesian Framework for Reinforcement Learning}, url = {http://citeseer.ist.psu.edu/strens00bayesian.html}, year = {2000} }

TY - CONF ID - citeulike:229813 L3 - citeulike-article-id:229813 TI - A Bayesian Framework for Reinforcement Learning T2 - Proc. 17th International Conf. on Machine Learning SP - 943 EP - 950 PB - Morgan Kaufmann, San Francisco, CA N2 - The reinforcement learning problem can be decomposed into two parallel types of inference: (i) estimating the parameters of a model for the underlying process; (ii) determining behavior which maximizes return under the estimated model. Following Dearden, Friedman and Andre (1999), it is proposed that the learning process estimates online the full posterior distribution over models. To determine behavior, a hypothesis is sampled from this distribution and the greedy policy with respect ... KW - bayesian KW - learning KW - reinforcement AU - Strens, Malcolm PY - 2000/// UR - http://citeseer.ist.psu.edu/strens00bayesian.html ER -