TicTacToeNN

This is a WPF application themed around a neural network that can play tic-tac-toe. The app provides two main functionalities: training the neural network and playing against it yourself. The used neural network is a Feedforward neural network that utilizes Q-Learning in its training process.

Features

• Interactive gameplay of tic-tac-toe
• Modifying the playboard size (available size range is [3,5])
• Controlling the training process via Start, Stop and Reset buttons
• Selecting a train partner for the network (random player, boring player, other neural network), as well as the amount of training rounds
• Seeing visual statistics of the training process

Screenshots

Start page view

Game mode

Training mode

Different playboard sizes

Technologies used

• .NET Framework 4.7.2
• C# 7.3
• Visual Studio 2017, refactored with Visual Studio 2022
• OxyPlot plotting library

Room for improvement

• Make it possible for the user to modify the network parameters
• Add the option to see the networks weights
• Add the ability to select the amount of cells needed for the win as well as the playboard size, e.g. 3 cells in a row on a 5x5 playboard
• Make the user interface more aesthetically pleasing

Acknowledgements

This project was completed as a coursework on machine learning named "Tic-Tac-Toe Neural Network" ("Нейронная сеть для игры в Крестики-Нолики"). The classes for the neural network and back propagation were based on the AForge.NET framework classes.