English

AsyncQVI: Asynchronous-Parallel Q-Value Iteration for Discounted Markov Decision Processes with Near-Optimal Sample Complexity

Optimization and Control 2020-02-25 v3 Machine Learning

Abstract

In this paper, we propose AsyncQVI, an asynchronous-parallel Q-value iteration for discounted Markov decision processes whose transition and reward can only be sampled through a generative model. Given such a problem with S|\mathcal{S}| states, A|\mathcal{A}| actions, and a discounted factor γ(0,1)\gamma\in(0,1), AsyncQVI uses memory of size O(S)\mathcal{O}(|\mathcal{S}|) and returns an ε\varepsilon-optimal policy with probability at least 1δ1-\delta using O~(SA(1γ)5ε2log(1δ))\tilde{\mathcal{O}}\big(\frac{|\mathcal{S}||\mathcal{A}|}{(1-\gamma)^5\varepsilon^2}\log(\frac{1}{\delta})\big) samples. AsyncQVI is also the first asynchronous-parallel algorithm for discounted Markov decision processes that has a sample complexity, which nearly matches the theoretical lower bound. The relatively low memory footprint and parallel ability make AsyncQVI suitable for large-scale applications. In numerical tests, we compare AsyncQVI with four sample-based value iteration methods. The results show that our algorithm is highly efficient and achieves linear parallel speedup.

Keywords

Cite

@article{arxiv.1812.00885,
  title  = {AsyncQVI: Asynchronous-Parallel Q-Value Iteration for Discounted Markov Decision Processes with Near-Optimal Sample Complexity},
  author = {Yibo Zeng and Fei Feng and Wotao Yin},
  journal= {arXiv preprint arXiv:1812.00885},
  year   = {2020}
}

Comments

Accepted by AISTATS 2020

R2 v1 2026-06-23T06:29:38.610Z