Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by jacquesm on Thursday March 10, 2016 @05:35AM
(#11258355)
With the amount of computing power they throw at this I doubt that's something even google could easily fund.
[ Parent
| on HN ]
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by malanj on Thursday March 10, 2016 @06:02AM
(#11258451)
Is running AlphaGo really that expensive? I get that training deep learning systems is very computationally expensive, but my understanding is that running them is orders of magnitude cheaper.
edit: table showing gains as CPUs are added - https://i.imgur.com/xxdWUtV.png
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by joosters on Thursday March 10, 2016 @06:05AM
(#11258463)
It's a huge cluster of machines.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by enneff on Thursday March 10, 2016 @06:13AM
(#11258491)
Got any numbers?
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by tim333 on Thursday March 10, 2016 @06:34AM
(#11258568)
The Economist says "The version playing against Mr Lee uses 1,920 standard processor chips and 280 special ones developed originally to produce graphics for video games"
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by kyrra on Thursday March 10, 2016 @11:23AM
(#11260083)
If you price by the GCE calculator[0] it's $1920 to rent 1920 CPU cores for 20 hours. This doesn't include GPU costs as they don't seem to have GPUs available on cloud, but I could see that easily doubling the costs.
[0] https://cloud.google.com/products/calculator/#id=f63f1fe0-56...
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by tim333 on Thursday March 10, 2016 @08:20PM
(#11263978)
Daresay the Economist is unclear whether they are talking chips or cores so it could be several times more cores.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by eru on Thursday March 10, 2016 @06:43AM
(#11258609)
So, that's actually not very expensive to rent for the duration of five games.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by Matt3o12_ on Thursday March 10, 2016 @07:09AM
(#11258697)
I don't know how long a game takes but I don't think google could afford letting anybody play, as stated in the original post.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by taneq on Thursday March 10, 2016 @09:23AM
(#11259323)
Charge $1000 per game and pay out $10,000 if the human wins. Could be a nice earner. ;)
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by Splines on Thursday March 10, 2016 @11:25AM
(#11260097)
I'll pay for two simultaneous sessions of AlphaGo please :). I'll even let it go first in one of them!
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by tim333 on Thursday March 10, 2016 @06:52PM
(#11263531)
It would be nice if they preserve a snap shot of Alphago as it was a the start of the Sedol games as a historic thing. Then if they open source it people could go play it too. They have mostly open sourced their code but not I think the learning data.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by eru on Friday March 11, 2016 @09:12PM
(#11270973)
They would only need to open the learned coefficients, not the data learned from.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by taneq on Thursday March 10, 2016 @08:35PM
(#11264054)
I like the way you think. :D
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by ProblemFactory on Thursday March 10, 2016 @12:16PM
(#11260448)
Clever, but they could easily detect this on their servers.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by bnegreve on Thursday March 10, 2016 @05:57AM
(#11258442)
But that would prove that they are not overfitting their opponent. Which would be nice.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by jacquesm on Thursday March 10, 2016 @05:59AM
(#11258447)
Not really. Even if it would only take a single dedicated GPU to win in real time against a human opponent you could not offer a service like that for free. If google massively overfits their opponent that equation still holds true so it's not really proof. It would only be proof if they switched off 10% or so of the array that powers their offering right now and it would lose from Lee Sedol consistently. For all you know they have a huge margin of error.
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by bnegreve on Thursday March 10, 2016 @06:04AM
(#11258460)
I suspect that training the AI is the most CPU intensive part.
If they would release the (trained) AI to everybody, that would prove that the training phase is general enough to beat any player, not just one.
Google doesn't have to pay for the CPU time. Ke Jie can find sponsors if he pretends he can beat AlphaGo.
(Edit: clarify)
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by eru on Thursday March 10, 2016 @06:44AM
(#11258615)
Alas, the software is tied to Google's platform. Just releasing as open source won't do. (That's why they didn't open source Google Reader. The source would have been both useless for running a Reader clone, and given away lots of trade secrets about Google infrastructure.)
Re:AlphaGo Can't Beat Me, Says Chinese Go Grandmaster Ke Jie
by jacquesm on Thursday March 10, 2016 @06:07AM
(#11258470)
I don't doubt there is an a-symmetry between the number of machines / GPUs they throw at the problem during a match and during the run-up to a match but even so they will have to have some margin of error if they expect to win in the first place and besides that whatever that pile of hardware is it, the infrastructure required to run it and the people involved are not free.