Dueling DQN

E98474

Deep Q-Network variant deep reinforcement learning algorithm value-based reinforcement learning method

Dueling DQN is a deep reinforcement learning algorithm that separates state-value and advantage estimations within its neural network architecture to improve learning efficiency and stability over standard DQN.

Try in SPARQL Jump to: Surface forms Statements Referenced by

All labels observed (4)

Label	Occurrences
Dueling DQN canonical	4
Dueling Deep Q-Network	1
Dueling Double DQN	1
Dueling Network Architectures for Deep Reinforcement Learning	1

Statements (44)

Predicate	Object
instanceOf	Deep Q-Network variant ⓘ deep reinforcement learning algorithm ⓘ value-based reinforcement learning method ⓘ
actionSpaceType	discrete ⓘ
aimsToImprove	learning efficiency ⓘ training stability ⓘ
basedOn	Q-learning ⓘ
citationVenue	ICML ⓘ surface form: Proceedings of the 33rd International Conference on Machine Learning
combinesStreamsToEstimate	Q-values ⓘ
commonlyEvaluatedOn	Atari 2600 ⓘ surface form: Atari 2600 games
controlType	off-policy ⓘ
domain	artificial intelligence ⓘ
especiallyHelpsWhen	many actions have similar value ⓘ only a few actions affect the value of the state ⓘ
extends	Atari deep Q-network ⓘ surface form: Deep Q-Network
field	reinforcement learning ⓘ
hasComponent	advantage stream ⓘ value stream ⓘ
hasFullName	Dueling DQN self-linksurface differs ⓘ surface form: Dueling Deep Q-Network
hasKeyIdea	decouple representation of state value from representation of advantages for each action ⓘ
implementedIn	DeepMind Atari agent ⓘ
improvesOver	Atari deep Q-network ⓘ surface form: standard DQN
influenced	Rainbow DQN ⓘ
introducedBy	Hado van Hasselt ⓘ Marc Lanctot ⓘ Matteo Hessel ⓘ Nando de Freitas ⓘ Tom Schaul ⓘ Ziyu Wang ⓘ
introducedInPaper	Dueling DQN self-linksurface differs ⓘ surface form: Dueling Network Architectures for Deep Reinforcement Learning
learningParadigm	model-free ⓘ
normalizesAdvantageStream	by subtracting mean advantage ⓘ
oftenCombinedWith	Double DQN ⓘ Prioritized Experience Replay DQN ⓘ surface form: Prioritized Experience Replay
publishedAtConference	ICML ⓘ surface form: ICML 2016
separatesEstimationOf	advantage function ⓘ state-value function ⓘ
sharesFeatureExtractor	between value and advantage streams ⓘ
usesFunctionApproximator	deep neural network ⓘ
usesLossFunction	temporal-difference loss ⓘ
usesOptimizationMethod	Adam optimizer ⓘ stochastic gradient descent ⓘ
usesTargetNetwork	yes ⓘ
yearIntroduced	2016 ⓘ

Referenced by (7)

Full triples — surface form annotated when it differs from this entity's canonical label.

OpenAI Baselines → implementsAlgorithm → Dueling DQN ⓘ

Atari deep Q-network → inspiredAlgorithm → Dueling DQN ⓘ

Dueling DQN → hasFullName → Dueling DQN self-linksurface differs ⓘ

this entity surface form: Dueling Deep Q-Network

Dueling DQN → introducedInPaper → Dueling DQN self-linksurface differs ⓘ

this entity surface form: Dueling Network Architectures for Deep Reinforcement Learning

Prioritized Experience Replay DQN → compatibleWith → Dueling DQN ⓘ

Double DQN → influenced → Dueling DQN ⓘ

this entity surface form: Dueling Double DQN

Rainbow DQN → improvesOver → Dueling DQN ⓘ