AlphaZero is not a scientific experiment but an industrial project with more or less advertisement for DeepMind.
So we should not expect a detailed description of their work.
Moreover if the expectations of Google about method of AlphaZero proves to be established they works has some aspect of national security, too.
So Alpha Zero was a hoax?
Moderators: hgm, Rebel, chrisw
-
- Posts: 3657
- Joined: Wed Nov 18, 2015 11:41 am
- Location: hungary
-
- Posts: 291
- Joined: Wed May 08, 2013 6:49 am
Re: So Alpha Zero was a hoax?
Even just reaching the level of TSCP is enough to convince me that this experiment works and can be further improved IMO.
-
- Posts: 47
- Joined: Mon Oct 31, 2016 9:45 pm
Re: So Alpha Zero was a hoax?
That is, in fact, exactly what I was saying (I even made a similar remark a while back in TCEC chat), but it seems that Bojun is too mindkilled to actually respond to my points. When a poster starts digging through your post history to look for ammunition against you as a person, instead of addressing your points directly, they have effectively conceded defeat.syzygy wrote:Isn't David just saying that in May 2017, when you made that statement, it sounded completely convincing? What you wrote there is almost exactly what I was still saying in October and November 2017 after the AlphaGo Zero results had become known. Unfortunately for me though, the AlphaZero results rather strongly suggest that I got it all wrong.noobpwnftw wrote:Oh, really? There is it when I said about the differences and possible chess usage, now I throw this back to you, where were you when there is no AlphaZero to brag around?David Xu wrote:Your claim that chess is somehow different from Go, likewise, would have been much more convincing had it been made before the AlphaZero results were published.
Based on this, looks like you have finally found your Messiah?
I am in fact completely aware of how traditional chess engines work. However, my knowledge of traditional chess engines has precisely zero bearing on the topic of whether a RL approach is superior, and that is the question at hand here. Bojun has done a very good job at avoiding making his position on this question explicit, but reading between the lines makes it very clear what he thinks the answer is.
Anyway, I don't recommend responding to him or Milos Stanisavljevic any further; it's not likely to be a productive discussion.