We call c(m) the m-step concentrability of a future-state distribution and call Cρ,µ the discountedaverage concentrability coefficient of the future-state distributions. The class of MDPs that satisfies this concentrability assumption is quite large, which is further discussed in Munos and Szepesvári [18]. If Xi, i = 1,...,N is an i.i.d. And when q = 1, N is used instead of N1. From the definition, one can esasily see that Nq,FX1:N N. Lemma A.2. (Single Iteration Error Bound) Let Vk and Vk+1 be the value functions of iteration kand k+1, and Vmax = rmax/(1 γ).

Batch reinforcement learning (RL) is important to apply RL algorithms to many high stakes tasks. Doing batch RL in a way that yields a reliable new policy in large domains is challenging: a new decision policy may visit states and actions outside the support of the batch data, and function approximation and optimization with limited samples can further increase the potential of learning policies with overly optimistic estimates of their future performance. Some recent approaches to address these concerns have shown promise, but can still be overly optimistic in their expected outcomes. Theoretical work that provides strong guarantees on the performance of the output policy relies on a strong concentrability assumption, which makes it unsuitable for cases where the ratio between state-action distributions of behavior policy and some candidate policies is large. This is because, in the traditional analysis, the error bound scales up with this ratio. We show that using \emph{pessimistic value estimates} in the low-data regions in Bellman optimality and evaluation back-up can yield more adaptive and stronger guarantees when the concentrability assumption does not hold. In certain settings, they can find the approximately best policy within the state-action space explored by the batch data, without requiring a priori assumptions of concentrability. We highlight the necessity of our pessimistic update and the limitations of previous algorithms and analyses by illustrative MDP examples and demonstrate an empirical comparison of our algorithm and other state-of-the-art batch RL baselines in standard benchmarks.

Batch reinforcement learning (RL) is important to apply RL algorithms to many high stakes tasks. Doing batch RL in a way that yields a reliable new policy in large domains is challenging: a new decision policy may visit states and actions outside the support of the batch data, and function approximation and optimization with limited samples can further increase the potential of learning policies with overly optimistic estimates of their future performance. Some recent approaches to address these concerns have shown promise, but can still be overly optimistic in their expected outcomes. Theoretical work that provides strong guarantees on the performance of the output policy relies on a strong concentrability assumption, which makes it unsuitable for cases where the ratio between state-action distributions of behavior policy and some candidate policies is large. This is because, in the traditional analysis, the error bound scales up with this ratio.