000190886 001__ 190886
000190886 005__ 20180317093332.0
000190886 037__ $$aCONF
000190886 245__ $$aCover Tree Bayesian Reinforcement Learning
000190886 269__ $$a2013
000190886 260__ $$aArxiv$$c2013
000190886 336__ $$aConference Papers
000190886 520__ $$aThis paper proposes an online tree-based Bayesian approach for reinforcement learning. For inference, we employ a generalised context tree model. This defines a distribution on multivariate Gaussian piecewise-linear models, which can be updated in closed form. The tree structure itself is constructed using the cover tree method, which remains efficient in high dimensional spaces. We combine the model with Thompson sampling and approximate dynamic programming to obtain effective exploration policies in unknown environments. The flexibility and computational simplicity of the model render it suitable for many reinforcement learning problems in continuous state spaces. We demonstrate this in an experimental comparison with least squares policy iteration.
000190886 700__ $$0246704$$aTziortziotis, Nikolaos$$g229875
000190886 700__ $$0245398$$aDimitrakakis, Christos$$g158533
000190886 700__ $$aBlekas, Konstantinos
000190886 7112_ $$aInternational Joint Conference on Artificial Intelligence, IJCAI 2013
000190886 909CO $$ooai:infoscience.tind.io:190886$$pIC$$pconf
000190886 909C0 $$0252184$$pLIA$$xU10406
000190886 917Z8 $$x158533
000190886 917Z8 $$x208605
000190886 917Z8 $$x208605
000190886 917Z8 $$x208605
000190886 937__ $$aEPFL-CONF-190886
000190886 973__ $$aEPFL$$rNON-REVIEWED$$sPUBLISHED
000190886 980__ $$aCONF