This repository is used to develop an AI for the Quoridor Game. In this part of the project, we try to optimize the AI.
- Activation functions :
- ReLU
- Leaky ReLU
- Hyperbolic tangent
- SWISH
- E-Greedy :
- Decreasing
- Bruit
- Soft-max
- Modification of the GUI :
- Can choose the value of the decreasing E-greedy and the speed of the decay rate
- Can choose which gloutonne strategy and mix Soft-max and E-greedy
- Can choose the activation function
- Can now train the IA and compare it with more than 1 wall
- Numpy,os,sys libraries
- Python3
- IA_partie2.py
- partie4.py
- utils_partie2.py
- Get the file tournoi.py and you will need :
- IA_partie2.py
- utils_partie2.py
- partie4.py
- Open your terminal and use the cmd :
- python3 tournoi.py nb_of_games_to_train_the_AI name_of_new_file.npz
- i.e : python3 tournoi.py 10000 my_new_AI.npz
Now you know everything you need to modify, use and enjoy this program ! 👌
This project has been made as a Year Project in my Computer Science freshman year at the University of Brussels. Credits to : Gwenaël Joret, Charlotte Nachtegael, Arnaud Pollaris, Cédric Ternon, Jérôme De Boeck.