Skip to content

A Python implementation of an Othello agent trained with temporal difference learning (TD-Lambda)

Notifications You must be signed in to change notification settings

ghassanachi/td-othello

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

74 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TD-Othello README

CS 701 Fall 2016 Seminar Project, Middlebury College

How to play (in othello/playOthello.py):
1. Load a neural network. These networks are loaded with the following inputs:
    - filename
    - training player
    - opponent training player
    - lambda value (9 for 0.9, 1 for 1.0)
2. Choose a mode to play with:
    - play0: Plays a game with two computer players in the command line
    - play1: Plays a game with one computer player and one human player in the command line
    - play2: Plays a game with two human players in the command line
    - playGui: Plays a game with one computer player and one human player in a Tkinter GUI
TD Lambda algorithm:

TD Lambda is a temporal difference learning algorithm designed by Richard Sutton. More info on the algorithm can be found in the paper in this repository or at this link: https://webdocs.cs.ualberta.ca/~sutton/papers/sutton-89.pdf

Known bugs:

With lambda < 1, the output values of the neural network approach 1 for all board states at some point during training. Performance decreases when this occurs.

About

A Python implementation of an Othello agent trained with temporal difference learning (TD-Lambda)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • TeX 68.3%
  • Python 31.7%