Dynamic Planning and Learning under Recovering Rewards

                    David Simchi-Levi 1 Zeyu Zheng 2 Feng Zhu 1

             Abstract                 immediately drops after it is pulled, and then gradually re-
    Motivated by emerging applications such as live-      covers if the arm is not pulled in the subsequent time periods.
    streaming e-commerce, promotions and recom-         This class of problems are motivated by emerging applica-
    mendations, we introduce a general class ...