Game-AI Forum

I have not kept up-to-date with this research topic, but I know it has been studied a lot. I found this with a quick Google search:
https://sites.google.com/site/dataefficientml/
Self-supervised learning is a related trend recently. You can probably find a lot more by searching the web.

Feel free to create your TORCS thread in the General Discussion (or here if you use tree search, but it is unlikely).

The successor to RARS is TORCS:
http://www.berniw.org/trb/
It is more complex, but also more interesting. I made a TORCS driver during my thesis, too. It was fun.

A way to get a recognition for a good machine-learning system is to win competitions. Some don't require that you disclose your code or method. Kaggle has competitions with big money prizes:
https://www.kaggle.com/competitions

I did not know this game, and it looks interesting. Since AlphaZero, there is little room for significant conceptual progress in perfect-information board games. But games with imperfect information are much more complex. Counterfactual Regret Minimization (CFR) seems to be an efficient approach in ...

Hi Daur, Thanks for your message. Regarding my personal feelings, I don't feel any frustration about the success of Alpha Go. I did receive quite a lot of recognition, and I have a dream job thanks to it. But I think I can understand your feelings. I was unemployed for one year after my PhD thesis, ...

Hi Daur, Welcome to the forum. This place has not been very alive for a while. If you wish to exchange with other Go programmers, I recommend the computer-go mailing list. Its activity has been very limited in the recent past, but many Go programmers are still reading it. The Computer Chess Club als...

Hi Christian,

Yes, I know Ludii. In fact I have been exchanging with the Ludii people on their forum:
https://ludii.games/forums/showthread.php?tid=87

Hi,

Yes, I am very interested.

Do you have a protocol for automatic play?

Rémi

Hi Alvaro, Because of hidden information, methods that work for board games usually don't work well for card games. The AlphaZero method itself may not work well, but other machine learning algorithms based on self play have been successfully applied to card games. The main approach for statistical ...