Skip to content

A reinforcement learning model Q-learning used in simple maze game.

License

Notifications You must be signed in to change notification settings

wendili-cs/Q-Learning_Maze

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Q-Learning_Maze

A reinforcement learning model Q-learning used in simple maze game.

Introduction

  • A training model on a simple maze game: blue square is the character; green square is the pickup which can increase, +1 point; then the red square is the trap that decrease, -1 point.

  • The model uses Q-learning, including main QValue Network and target QValue Network.

  • The final training on test may sometimes gets stucked, please try again before I fix it. I have fix it at 2017.12.03, and add another type of object, and a mode that randomly gives number of each kind of objexts.

pic

pic


How to use

  • Choose to train or to test by setting the toTest and if you want to test, set testSteps to set the action steps that character allow to take.

About

A reinforcement learning model Q-learning used in simple maze game.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages