Thanks to visit codestin.com
Credit goes to github.com

Skip to content

lwintjen/Quoridor

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

QUORIDOR GAME - LEARNING BY REINFORCEMENT

This repository is used to develop an AI for the Quoridor Game. In this part of the project, we try to optimize the AI.

The following updates have been added in the code :

  • Activation functions :
    1. ReLU
    2. Leaky ReLU
    3. Hyperbolic tangent
    4. SWISH
  • E-Greedy :
    1. Decreasing
    2. Bruit
  • Soft-max
  • Modification of the GUI :
    1. Can choose the value of the decreasing E-greedy and the speed of the decay rate
    2. Can choose which gloutonne strategy and mix Soft-max and E-greedy
    3. Can choose the activation function
  • Can now train the IA and compare it with more than 1 wall

What do you need to run the program :

  • Numpy,os,sys libraries
  • Python3
  • IA_partie2.py
  • partie4.py
  • utils_partie2.py

If you want to use the program without a GUI :

  • Get the file tournoi.py and you will need :
    1. IA_partie2.py
    2. utils_partie2.py
    3. partie4.py
  • Open your terminal and use the cmd :
    1. python3 tournoi.py nb_of_games_to_train_the_AI name_of_new_file.npz
    2. 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.

About

Reinforcement learning using a neural network for the board game Quoridor

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages