QUOTA: The Quantile Option Architecture for Reinforcement Learning
DOI:
https://doi.org/10.1609/aaai.v33i01.33015797Abstract
In this paper, we propose the Quantile Option Architecture (QUOTA) for exploration based on recent advances in distributional reinforcement learning (RL). In QUOTA, decision making is based on quantiles of a value distribution, not only the mean. QUOTA provides a new dimension for exploration via making use of both optimism and pessimism of a value distribution. We demonstrate the performance advantage of QUOTA in both challenging video games and physical robot simulators.
Downloads
Published
2019-07-17
How to Cite
Zhang, S., & Yao, H. (2019). QUOTA: The Quantile Option Architecture for Reinforcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 5797-5804. https://doi.org/10.1609/aaai.v33i01.33015797
Issue
Section
AAAI Technical Track: Machine Learning