[go: up one dir, main page]

Skip to content

An AI program for Game 2048 based on temporal difference learning, N-Tuple Networks and expectimax search

Notifications You must be signed in to change notification settings

ccchang1023/CGI-2048

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CGI-2048

CGI-2048 is an AI program for Game 2048. Developed by CGI-Lab(http://www.aigames.nctu.edu.tw/) Kun-Hao Yeh, Chao-Chin Liang, Chia-Chuan Chang in 2014~2015

The performance of CGI-2048 is still improving. The table shows the performances of the early CGI-2048 (the one for the tournament), the latest improved CGI-2048, and Xificurk's program. To the best of our knowledge, Xificurk's outperformed all the known 2048 programs at the time before Octobor 2014. In November 2014, our latest CGI-2048 is able to perform better than Xificurk's. More importantly, ours runs about 300 times faster.

Our algorithm is based on the following two papers: [1] Szubert, M. amd Jaskowaski, W., “Temporal Difference Learning of N-tuple Networks for the Game 2048”, IEEE CIG 2014 conference, August 2014.

[2] I-Chen Wu, Kun-Hao Yeh, Chao-Chin Liang, Chia-Chuan Chang, and Han Chiang, "Multi-Stage Temporal Difference Learning for 2048", the 2014 Conference on Technologies and Applications of Artificial Intelligence (TAAI 2014), Taipei, Taiwan, Novemeber 2014.

About

An AI program for Game 2048 based on temporal difference learning, N-Tuple Networks and expectimax search

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published