GPTKB
Browse
Query
Compare
Download
Publications
Contributors
Search
Dueling Network Architectures
URI:
https://gptkb.org/entity/Dueling_Network_Architectures
GPTKB entity
Statements (34)
Predicate
Object
gptkbp:instanceOf
Neural network architecture
gptkbp:architecture
Deep neural network
gptkbp:arXivID
1511.06581
gptkbp:benefit
Better policy evaluation
Improved learning efficiency
gptkbp:citation
5000+
gptkbp:combines
Value and advantage streams to produce Q-values
gptkbp:component
Advantage stream
Value stream
gptkbp:contribution
Separates estimation of value and advantage in Q-networks
https://www.w3.org/2000/01/rdf-schema#label
Dueling Network Architectures
gptkbp:improves
Deep Q-Networks
gptkbp:introduced
gptkb:Nando_de_Freitas
gptkb:Hado_van_Hasselt
gptkb:Marc_Lanctot
gptkb:Matteo_Hessel
gptkb:Tom_Schaul
gptkb:Ziyu_Wang
gptkbp:notablePublication
gptkb:Dueling_Network_Architectures_for_Deep_Reinforcement_Learning
gptkbp:openSource
gptkb:OpenAI_Baselines
gptkb:Stable_Baselines
gptkb:TensorFlow_Agents
gptkbp:publicationYear
2016
gptkbp:publishedIn
gptkb:ICML_2016
gptkbp:relatedTo
gptkb:Double_DQN
gptkb:Prioritized_Experience_Replay
gptkbp:separates
Advantage function
State-value function
gptkbp:usedFor
gptkb:Q-learning
Value-based reinforcement learning
gptkbp:usedIn
Deep reinforcement learning
Atari 2600 benchmarks
gptkbp:bfsParent
gptkb:Rainbow_DQN
gptkbp:bfsLayer
7