README.md

Connect4 AI

Python bot to play connect4, uses Q-Learning based on the Greedy Epsilon Bellman Equation.

https://youtube.com/shorts/njdJeu95p6s?si=jI39STsb80ijHa8h

Installation

1. Unzip q_table_shelf.db.zip and place the three files in root.
2. Install numpy

pip install numpy

Usage

• To play a best of 3 against the AI:

   python connect4.py

• To train the model:

   python train.py

  By default this will train the model for 60,000 episodes, the model in q_table_shelf.db has been trained for 200,000 episodes/10 hours.

Information

The gameboard is a 2D list and is represented in the following image, positions to drop a COIN are from 1 to 7, player one is 7, the AI is player two or 5. 0 represents an empty space.

References

• Trust me bro
• An AI agent learns to play tic-tac-toe (part 3): training a Q-learning RL agent
• Notebook for Topic 08 Video - Q-Learning - A Complete Example in Python.ipynb