English

Variance-reduced $Q$-learning is minimax optimal

Machine Learning 2019-08-09 v2 Optimization and Control Machine Learning

Abstract

We introduce and analyze a form of variance-reduced QQ-learning. For γ\gamma-discounted MDPs with finite state space X\mathcal{X} and action space U\mathcal{U}, we prove that it yields an ϵ\epsilon-accurate estimate of the optimal QQ-function in the \ell_\infty-norm using O((Dϵ2(1γ)3)  log(D(1γ)))\mathcal{O} \left(\left(\frac{D}{ \epsilon^2 (1-\gamma)^3} \right) \; \log \left( \frac{D}{(1-\gamma)} \right) \right) samples, where D=X×UD = |\mathcal{X}| \times |\mathcal{U}|. This guarantee matches known minimax lower bounds up to a logarithmic factor in the discount complexity. In contrast, our past work shows that ordinary QQ-learning has worst-case quartic scaling in the discount complexity.

Cite

@article{arxiv.1906.04697,
  title  = {Variance-reduced $Q$-learning is minimax optimal},
  author = {Martin J. Wainwright},
  journal= {arXiv preprint arXiv:1906.04697},
  year   = {2019}
}

Comments

Update from v1: new Proposition 1 on minimax optimality; updated referencing and discussion of related work

R2 v1 2026-06-23T09:50:33.390Z