Monte-Carlo Snake
A simple reinforcement learning (snake) game that I wrote for fun while exploring RL during my Masters. The learning algorithm of the snake is based on Monte-Carlo using every visit update. More details can be found in the code!
The project is composed of two main scripts:
- scripts/snake.py - Contains the snake logic (Monte-carlo, Q-learning, train/test loop). In order to train and test your algorithm please change the relevant parameters under the main section.
- scripts/SnakeGame.py - The actual snake game - (based on OpenAI Gym API https://gym.openai.com/envs/#classic_control).
Feel free to add you own algorithm and test it using the SnakeGame class in SnakeGame.py!
Enjoy!
