Inducing Partially Observable Markov Decision Processes


Michael L. Littman ;
Proceedings of the Eleventh International Conference on Grammatical Inference, PMLR 21:145-148, 2012.


The partially observable Markov decision process (POMDP) model plays an important role in the field of reinforcement learning. It captures the problem of decision making when some important features of the environment are not visible to the decision maker. A number of approaches have been proposed for inducing POMDP models from data, a problem that has important parallels with grammar induction.

Related Material