> ML_LITERATURE // MNIH-2015-HUMAN-LEVEL-CONTROL-THROUGH-DEEP-REINFORCEMENT-LEARNING-DQN_v1.0
Human-level control through deep reinforcement learning (Nature DQN)
Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A. Rusu, Joel Veness, Marc G. Bellemare, Alex Graves, Martin Riedmiller, Andreas K. Fidjeland, Georg Ostrovski, Stig Petersen, Charles Beattie, Sadik Sadik, Anton Antonoglou, Helen King, Dharshan Kumaran, Daan Wierstra, Shane Legg, Demis Hassabis · Nature (2015)
seminal-architecture2015foundationalthirdPartyReproduced
Principal Contribution
Stabilized deep Q-learning via experience replay buffers and separate frozen target networks, achieving superhuman performance across 49 Atari 2600 games directly from raw pixels.
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