A Cramér Distance perspective on Quantile Regression based Distributional Reinforcement Learning
Alix Lheritier, Nicolas Bondoux
Abstract
Distributional reinforcement learning (DRL) extends the value-based approach by approximating the full distribution over future returns instead of the mean only, providing a richer signal that leads to improved performances. Quantile Regression (QR)-based methods like QR-DQN project arbitrary distributions into a parametric subset of staircase distributions by minimizing the 1-Wasserstein distance. However, due to biases in the gradients, the quantile regression loss is used instead for training, guaranteeing the same minimizer and enjoying unbiased gradients. Non-crossing constraints on the quantiles have been shown to improve the performance of QR-DQN for uncertainty-based exploration strategies. The contribution of this work is in the setting of fixed quantile levels and is twofold. First, we prove that the Cramer distance yields a projection that coincides with the 1-Wasserstein one and that, under non-crossing constraints, the squared Cramer and the quantile regression losses yield collinear gradients, shedding light on the connection between these important elements of DRL. Second, we propose a low complexity algorithm to compute the Cramer distance.
BibTeX
@InProceedings{pmlr-v151-lheritier22a,
title = { A Cramér Distance perspective on Quantile Regression based Distributional Reinforcement Learning },
author = {Lheritier, Alix and Bondoux, Nicolas},
booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics},
pages = {5774--5789},
year = {2022},
editor = {Camps-Valls, Gustau and Ruiz, Francisco J. R. and Valera, Isabel},
volume = {151},
series = {Proceedings of Machine Learning Research},
month = {28--30 Mar},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v151/lheritier22a/lheritier22a.pdf},
url = {https://proceedings.mlr.press/v151/lheritier22a.html},
abstract = { Distributional reinforcement learning (DRL) extends the value-based approach by approximating the full distribution over future returns instead of the mean only, providing a richer signal that leads to improved performances. Quantile Regression (QR)-based methods like QR-DQN project arbitrary distributions into a parametric subset of staircase distributions by minimizing the 1-Wasserstein distance. However, due to biases in the gradients, the quantile regression loss is used instead for training, guaranteeing the same minimizer and enjoying unbiased gradients. Non-crossing constraints on the quantiles have been shown to improve the performance of QR-DQN for uncertainty-based exploration strategies. The contribution of this work is in the setting of fixed quantile levels and is twofold. First, we prove that the Cramer distance yields a projection that coincides with the 1-Wasserstein one and that, under non-crossing constraints, the squared Cramer and the quantile regression losses yield collinear gradients, shedding light on the connection between these important elements of DRL. Second, we propose a low complexity algorithm to compute the Cramer distance. }
}