The aim of this project is to implement a Q-learning and a deep Q-learning algorithm in order to play TicTacToe. The exact framing of the project can be found in the file "Project_instruction.pdf".
Our results can be found in "Report.pdf", and can be reproduced by running "Example.ipynb".
Here are two images extracted from "Report.pdf" that show the performances of our algorithms.
The first one show the performances of our Q-learning implementation:
And the second one show the performances of our deep Q-learning implementation:
This project was realised in the scope of the class "Artificial neural networks/reinforcement learning" (CS-456) thaugt by Gerstner Wolfram.