Agent
September 18, 2019 ยท View on GitHub
As of September 18, 2019.
Overview
Agents are implemented by wrapping Keras-RL API. Each file in this directory is a different type of reinforcement learning algo.
DQN Architecture
- Dueling architecture
- Double Q-learning
- Experience replay
Neural Networks
CNN
- 3x convolutional layers
- 1x dense mlp
MLP
- 2x Dense mlp