Skip to content

Heuristic evaluation of Othello board by multi-layer perceptron for effective 1-ply tree search (the more keywords the merrier)

Notifications You must be signed in to change notification settings

TheoVerhelst/SimpleOthelloNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 

Repository files navigation

SimpleOthelloNet

What's this?

This is a simple attempt to use a neural network to learn how to play the Othello game. It turns out that the neural network beats a minimax player at depth of 6 with a good heuristic designed by T. Yoshioka (see the doc for further references).

How does it work?

The network is trained by modifying its weights using temporal difference learning. It plays against a copy of itself, and takes the raw board as input. After a few 45,000 games, it becomes a decent player. See the doc for further explanations.

How can I try it?

It's pure java, with only the help of Neuroph library. Once you set up the project in your IDE and imported Neuroph, just run it. By default, the main function trains a neural network using some parameters, and save them to a file. You can edit it to do whatever you want.

About

Heuristic evaluation of Othello board by multi-layer perceptron for effective 1-ply tree search (the more keywords the merrier)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages