Welcome to the Connect4 game agents repository!
This repository contains two implementations of game agents for the popular game Connect4.
The agents use two reinforcement learning algorithms: Monte Carlo tree search and Q learning.
To use the agents, you will need to download the all the files and run them, you will have to install pre-requiste libraies for the same.
Pre-requisites:
Python 3.6+
NumPy
GZip
Pickle
PyGame
The Monte Carlo tree search agent uses a tree-based search algorithm to explore the game's decision space and select the best moves.
The Q learning agent approximate the Q function and learn from experience.
To train the Q learning agent, you can play it against the MCTS agent, and the values are stored in QTable. The agent will use this data to learn the optimal policy for playing Connect4.
Overall, these agents provide a fun and exciting way to explore the capabilities of reinforcement learning algorithms in the context of a popular game.