A personal and educational deep reinforcement learning project in Python using TensorFlow and Keras. Includes a custom DQN agent library and 4 examples covering Q-Learning and DQN.
The dqn module is a custom implementation of a Deep Q-Network (DQN) agent for reinforcement learning tasks. It includes utilities for experience replay, exploration policies, and Atari-specific preprocessing. This project is designed for educational purposes and demonstrates the application of reinforcement learning algorithms in various environments.
A simple implementation of Q-Learning for the Taxi-v3 environment using NumPy. Demonstrates tabular Q-Learning with exploration and exploitation strategies.
A DQN implementation for the CartPole-v1 environment using TensorFlow/Keras. Includes a simple feedforward neural network and experience replay.
A DQN agent trained to play Atari Breakout. Uses convolutional neural networks, frame stacking, and experience replay.
An advanced DQN implementation for Atari Breakout using vectorized environments to speed up training. Includes prioritized experience replay and parallel environment execution.
To install the dqn module, clone this repository and run the following command in the root directory:
pip install .This will install the dqn module along with its dependencies specified in requirements.txt.
Each example script can be run directly from the command line. For example:
python 01-taxi-qlearning.pyEnsure that the required dependencies are installed and that the dqn module is properly installed before running the examples.
This project is licensed under the MIT License. See the LICENSE file for details.