stay a little random O

stay a little random
Oh good. I’ve been waiting for someone (well, some predictive mechanism) to beat Go. It’s still not an all-purpose solution yet – it required several systems cobbled together for it to work, but, it succeeded in its task at beating some Go champions consistently.

Go is interesting because the “space of possible moves” is huge compared to chess, so it’s kinda impractical to have it try random moves and see what’s successful. I mean that works, but it would take a long time.

So, they put together two layers of processes it seems. At one level it tries to predict potential outcomes using the whole board, and at another level it predicts the best next possible moves.

So, out to in + in to out at the same time: big picture fuzzy pattern matching combined with move-by-move logic.

It’s customized for the game of course but took advantage of the same algorithms used by Google images that predicted eyes in everything.

http://www.technologyreview.com/news/546066/googles-ai-masters-the-game-of-go-a-decade-earlier-than-expected/

It’s an achievement and a good one. Don’t worry about mind reading though; that’s only true if you stay p red i k t a b…….l .. ____e…

Leave a comment

Your email address will not be published. Required fields are marked *


+ 8 = fourteen

Leave a Reply