English

Value-aware Importance Weighting for Off-policy Reinforcement Learning

Machine Learning 2023-06-28 v1 Artificial Intelligence

Abstract

Importance sampling is a central idea underlying off-policy prediction in reinforcement learning. It provides a strategy for re-weighting samples from a distribution to obtain unbiased estimates under another distribution. However, importance sampling weights tend to exhibit extreme variance, often leading to stability issues in practice. In this work, we consider a broader class of importance weights to correct samples in off-policy learning. We propose the use of value-aware importance weights\textit{value-aware importance weights} which take into account the sample space to provide lower variance, but still unbiased, estimates under a target distribution. We derive how such weights can be computed, and detail key properties of the resulting importance weights. We then extend several reinforcement learning prediction algorithms to the off-policy setting with these weights, and evaluate them empirically.

Keywords

Cite

@article{arxiv.2306.15625,
  title  = {Value-aware Importance Weighting for Off-policy Reinforcement Learning},
  author = {Kristopher De Asis and Eric Graves and Richard S. Sutton},
  journal= {arXiv preprint arXiv:2306.15625},
  year   = {2023}
}

Comments

CoLLAs 2023

R2 v1 2026-06-28T11:15:54.655Z