QR-DQN
A distributional reinforcement learning algorithm that learns the return distribution through quantile regression. It belongs to the class of quantile methods.
Timeline (last 90 days)
QR-DQN, proposed by Dabney et al. in 2018, uses quantile regression to learn the quantiles of the return distribution, directly parameterizing the quantiles and avoiding projection error.
The paper's quantile fixed-point estimator can be viewed as the theoretical counterpart of QR-DQN, and its rate analysis explains why QR-DQN performs well in practice.
All Facts (2)
QR-DQN, proposed by Dabney et al. in 2018, uses quantile regression to learn the quantiles of the return distribution, directly parameterizing the quantiles and avoiding projection error.
50%UnverifiedThe paper's quantile fixed-point estimator can be viewed as the theoretical counterpart of QR-DQN, and its rate analysis explains why QR-DQN performs well in practice.
50%