Distributional RL

GPTKB entity

Statements (27)
Predicate Object
gptkbp:instanceOf Reinforcement Learning Method
gptkbp:advantage Captures risk and uncertainty
More robust learning
gptkbp:application gptkb:robot
gptkb:Atari_Games
Finance
gptkbp:contrastsWith Expected Value RL
gptkbp:enables Improved performance
Better exploration
Uncertainty estimation
gptkbp:field gptkb:Machine_Learning
gptkb:artificial_intelligence
gptkb:Reinforcement_Learning
gptkbp:focusesOn Learning the distribution of returns
gptkbp:hasModel Random return distribution
https://www.w3.org/2000/01/rdf-schema#label Distributional RL
gptkbp:improves Value-based RL methods
gptkbp:introduced gptkb:Marc_G._Bellemare
gptkbp:introducedIn 2017
gptkbp:key A Distributional Perspective on Reinforcement Learning
gptkbp:relatedTo C51 Algorithm
Implicit Quantile Networks
Quantile Regression DQN
gptkbp:usedIn Deep RL Algorithms
gptkbp:uses Probability distributions
gptkbp:bfsParent gptkb:Rainbow_DQN
gptkbp:bfsLayer 7