https://github.com/johnygomez/atarirl
Atari games learning using reinforcement learning in Python
https://github.com/johnygomez/atarirl
Last synced: 3 months ago
JSON representation
Atari games learning using reinforcement learning in Python
- Host: GitHub
- URL: https://github.com/johnygomez/atarirl
- Owner: johnygomez
- License: mit
- Created: 2016-05-20T18:19:03.000Z (about 9 years ago)
- Default Branch: master
- Last Pushed: 2017-09-15T15:40:26.000Z (over 7 years ago)
- Last Synced: 2025-01-12T12:18:12.896Z (4 months ago)
- Language: Python
- Size: 5.05 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Atari Reinforcement Learning
Aim of this project is to implement simple learning mechanism for Atari 2600 games on ALE Platform ([Arcade Learning Environment](https://github.com/mgbellemare/Arcade-Learning-Environment)). As a reinforcement learning technique we use **NFQ Learning** (see next).
### Main Principles of learning:
* First of all, input image of a game is read by ALE interface (lib/ImgProc)
* The input image is encoded using Autoencoder in [lib/Autoencoder/Encoder.py](lib/Autoencoder/Encoder.py)
* The extracted features from encoder (and also reward, action selected.. = transition data) are pushed into the [NFQ module](lib/NFQ/NFQ.py) which saves these transition as a training data for Neural Network learning
* When enough training data are acquired, the NFQ trains the Q-function, which is used as a predictor of the most suitable move
*Find out more about NFQ at [link](http://ml.informatik.uni-freiburg.de/_media/publications/rieecml05.pdf)*.
### Dependencies
* [Pip](https://pypi.python.org/pypi/pip)
* [Numpy](http://www.numpy.org/)
* [Theano](http://deeplearning.net/software/theano/)
* [Keras](http://keras.io/)
* [ALE](https://github.com/mgbellemare/Arcade-Learning-Environment)
### Quick Start
```
make init
```
which installs all dependencies
```
cd example/
python PongAgent.py
```
### Pong game example
To run a simple game example, you first need to get a game binary (you can find one in [example/](example/) folder).
The next step is to train the encoder [lib/Autoencoder/Encoder.py](lib/Autoencoder/Encoder.py) for your game using train method.
When the encoder is train well enough, you can use this pre-trained NN within initialization of [AleAgent](lib/AleAgent.py) after the first required parameter, which is the Game ROM.
[AleAgent](lib/AleAgent.py) contains 2 more methods, which are `train()` and `play()` which first train the Q-function for prediction and then, can play the game using trained network.