> ML_LITERATURE // WANG-2016-DUELING-NETWORK-ARCHITECTURES-DEEP-REINFORCEMENT-LEARNING_v1.0
Dueling Network Architectures for Deep Reinforcement Learning (Dueling DQN)
Ziyu Wang, Tom Schaul, Matteo Hessel, Hado van Hasselt, Marc Lanctot, Nando de Freitas · International Conference on Machine Learning (ICML) (2016)
seminal-architecture2016foundationalthirdPartyReproduced
Principal Contribution
Decomposed the state-action value Q(s,a) into two separate streams: a state value estimator V(s) and state-dependent advantage estimator A(s,a).
Operational Relevance
Serves as qualified theoretical and systems foundation for task-reinforcement-learning.
Assumptions
- Markovian state dynamics and stationary reward functions hold in target evaluation environments
Limitations
- Sample efficiency, exploration stability, and real-world sim-to-real transfer gaps require specialized tuning