English

Exponential Lower Bounds for Batch Reinforcement Learning: Batch RL can be Exponentially Harder than Online RL

Machine Learning 2021-06-22 v4 Artificial Intelligence

Abstract

Several practical applications of reinforcement learning involve an agent learning from past data without the possibility of further exploration. Often these applications require us to 1) identify a near optimal policy or to 2) estimate the value of a target policy. For both tasks we derive \emph{exponential} information-theoretic lower bounds in discounted infinite horizon MDPs with a linear function representation for the action value function even if 1) \emph{realizability} holds, 2) the batch algorithm observes the exact reward and transition \emph{functions}, and 3) the batch algorithm is given the \emph{best} a priori data distribution for the problem class. Our work introduces a new `oracle + batch algorithm' framework to prove lower bounds that hold for every distribution. The work shows an exponential separation between batch and online reinforcement learning.

Keywords

Cite

@article{arxiv.2012.08005,
  title  = {Exponential Lower Bounds for Batch Reinforcement Learning: Batch RL can be Exponentially Harder than Online RL},
  author = {Andrea Zanette},
  journal= {arXiv preprint arXiv:2012.08005},
  year   = {2021}
}

Comments

Accepted to ICML 2021 as long talk

R2 v1 2026-06-23T20:58:26.606Z