Ivan Sorokin, Alexey Seleznev, Mikhail Pavlov, Aleksandr Fedorov, Anastasiia Ignateva
A deep learning approach to reinforcement learning led to a general learner able to train on visual input to play a variety of arcade games at the human and superhuman levels. Its creators at the Google DeepMind's team called the approach: Deep Q-Network (DQN). We present an extension of DQN by "soft" and "hard" attention mechanisms. Tests of the proposed Deep Attention Recurrent Q-Network (DARQN) algorithm on multiple Atari 2600 games show level of performance superior to that of DQN. Moreover, built-in attention mechanisms allow a direct online monitoring of the training process by highlighting the regions of the game screen the agent is focusing on when making decisions.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Atari Games | Atari 2600 Tutankham | Score | 197 | DARQN soft |
| Atari Games | Atari 2600 Breakout | Score | 20 | DARQN hard |
| Atari Games | Atari 2600 Gopher | Score | 5356 | DARQN soft |
| Atari Games | Atari 2600 Space Invaders | Score | 650 | DARQN soft |
| Atari Games | Atari 2600 Seaquest | Score | 7263 | DARQN soft |
| Video Games | Atari 2600 Tutankham | Score | 197 | DARQN soft |
| Video Games | Atari 2600 Breakout | Score | 20 | DARQN hard |
| Video Games | Atari 2600 Gopher | Score | 5356 | DARQN soft |
| Video Games | Atari 2600 Space Invaders | Score | 650 | DARQN soft |
| Video Games | Atari 2600 Seaquest | Score | 7263 | DARQN soft |