How Alpha Go works

[peti29]

It's been another longer break for me with Go, so I haven't been playing games or following anything Go related until recently.

But now I learned that Alpha Go has beaten Ke Jie a few days ago. And then when I watched some Nick Sibicky videos he kept referring to Alpha Go style... I got the feeling that Alpha Go became very influential.

I did some research and now I'm completely mesmerized by this phenomenon! I've yet to watch the commented games of Alpha Go vs Ke Jie and Lee Sedol. But what's even more interesting for me is how Alpha Go works!

I remember how disappointed I was at University some 15 years ago when at the AI class we only learned about A* tree search and the like. That's not artificial intelligence in my book, that's just dumbly calculating all (well, most) possible outcomes.

But then there is deep neural networks now! That's something! That's some real intelligence. I've yet to do a lot more research, but I already found some pretty good videos that explain what Alpha Go is doing:

https://www.youtube.com/watch?v=derC33ODrME
https://www.youtube.com/watch?v=b9H9AtbxpPM

If you know similarly easily understandable videos about Alpha Go or deep neural networks, please share!

Edit: This is also very interesting (co-founder of Deep Mind talks):
https://www.youtube.com/watch?v=qRSAQWCh_yM
(for some reason the video is repeated 2 times)

[Kirby]

Andrew Ng's course on machine learning is kind of interesting.

https://www.coursera.org/learn/machine-learning

[Matthew Lowton]

Interesting, thanks for the link! That's really something! That's some kinda real intelligence.

[pookpooi]

Don't forget this latest clip in DeepMind channel

https://www.youtube.com/watch?v=Bm7zah_LrmE