Corpus ID: 216036114

# Almost Optimal Model-Free Reinforcement Learning via Reference-Advantage Decomposition

@article{Zhang2020AlmostOM,
title={Almost Optimal Model-Free Reinforcement Learning via Reference-Advantage Decomposition},
author={Zihan Zhang and Y. Zhou and Xiangyang Ji},
journal={ArXiv},
year={2020},
volume={abs/2004.10019}
}
• Published 2020
• Computer Science, Mathematics
• ArXiv
We study the reinforcement learning problem in the setting of finite-horizon episodic Markov Decision Processes (MDPs) with $S$ states, $A$ actions, and episode length $H$. We propose a model-free algorithm UCB-Advantage and prove that it achieves $\tilde{O}(\sqrt{H^2SAT})$ regret where $T = KH$ and $K$ is the number of episodes to play. Our regret bound improves upon the results of [Jin et al., 2018] and matches the best known model-based algorithms as well as the information theoretic lower… Expand
30 Citations

#### References

SHOWING 1-10 OF 29 REFERENCES