On The Statistical Complexity of Offline Decision-Making

Thanh Nguyen-Tang; Raman Arora

ICML 2024poster2 citations

On The Statistical Complexity of Offline Decision-Making

Abstract

We study the statistical complexity of offline decision-making with function approximation, establishing (near) minimax-optimal rates for stochastic contextual bandits and Markov decision processes. The performance limits are captured by the pseudo-dimension of the (value) function class and a new characterization of the behavior policy that *strictly* subsumes all the previous notions of data coverage in the offline decision-making literature. In addition, we seek to understand the benefits of using offline data in online decision-making and show nearly minimax-optimal rates in a wide range of regimes.

BibTeX

@inproceedings{
nguyen-tang2024on,
title={On The Statistical Complexity of Offline Decision-Making},
author={Thanh Nguyen-Tang and Raman Arora},
booktitle={Forty-first International Conference on Machine Learning},
year={2024},
url={https://openreview.net/forum?id=dYDPcx78tm}
}