作者: Peng Zhao, Long-Fei Li, Zhi-Hua Zhou
摘要
We investigate online Markov Decision Processes (MDPs) with adversarially changing loss functions and known transitions. We choose