TY  - RPRT
TI  - Exponential Lower Bounds for Batch Reinforcement Learning: Batch RL can be Exponentially Harder than Online RL
AU  - Andrea Zanette
PY  - 2021
UR  - https://arxiv.org/abs/2012.08005
ID  - 2012.08005
ER  -