Tic-Tac-Toe AI

Welcome to the Tic-Tac-Toe AI project! This repository aims to develop an intelligent agent capable of playing Tic-Tac-Toe and predicting the next move to play. The project involves data collection, model training, and evaluation to create a robust AI that can compete against human players or other AI opponents.

Table of Contents

• Introduction
• Features
• Installation
• Usage
• Data Simulation

Introduction

Tic-Tac-Toe is a simple yet classic game played on a 3x3 grid. Players take turns marking a cell in the grid with their symbol (x or o), aiming to place three of their marks in a horizontal, vertical, or diagonal row. Despite its simplicity, Tic-Tac-Toe serves as an excellent starting point for developing and testing AI algorithms.

This project is designed to create an AI that can (for now):

• Simulate games and collect data.
• Train on game data to predict the best next move.
• Train with the hand to understand what really do a training model

Features

• Game Simulation: Automatically simulate and record multiple games of Tic-Tac-Toe.
• Data Processing: Convert game records into tensors for machine learning models.
• Debugging function for each step: Understand each layer of the model with vizualisation of tensor and his values

Installation

Prerequisites

• Python 3.8+
• PyTorch
• pandas
• csv
• random

Clone the Repository

git clone https://github.com/ramosleandre/Tic_Tac_Toe_AI.git
cd Tic_Tac_Toe_AI

Install Dependencies

pip install -r requirements.txt

Usage

Data Simulation

Simulate games and generate a CSV file with game states and moves.

import csv
import random

# Add the simulate_games function here
# Your simulation code from above

# Run the simulation for N games
simulate_games(1000)

Convert CSV to Tensors

Convert the simulated game data into tensors for model training. choose your dataset and for debug, the line at debug.

tensor_converter = CSVToTensor('./Datasets/tic_tac_toe_10_games.csv')
tensor_converter.csv_to_tensor(position)
print(f"Input : {tensor_converter.game_tensor[position]}")
print(f"Output : {tensor_converter.prediction_tensor[position]}")

Output :

Input : tensor([1., 0., 1., 2., 0., 0., 0., 0., 0.])
Output : tensor([0., 2., 0., 0., 0., 0., 0., 0., 0.])

Convert you tensor (position in your CSV) in a vizualizer

tensor_converter.tensor_to_view(position)

Output :

Current Game State:
x | O | x
---------
o |   |  
---------
  |   |