Reinforcement Learning for Atari (2015)
Deep Attention Recurrent Q-Network
This work has been presented at NeurIPS Deep Reinforcement Learning Workshop 2015.
DeepHack.Game First Place Solution
The hackathon DeepHack.Game had a goal the improve the performance of the Google DeepMindās Q-learning algorithm DQN. The DQN is known for its ability to play Atari 2600 games at human and superhuman levels from pixel input.