Q-Learning Kivy Tic-Tac-Toe
This project acts as my introduction to the concept of training an AI with Reinforcement. As TicTacToe isn't too complex, I was able to achieve very good results without having to use any neural network, but just a state hashtable with a learned score for each state (Q-learning). The states are represented by trinari numbers.
For graphics I used the Kivy-Python graphics library.
The original script is in python, but I have compiled it, together with all the necesseray bundles into a Windows Executable file which you can download to play. (It can also run sandboxed, if you don't trust the program)
Download Link (via Google Drive):
TicTacToe.exe