It is currently Fri Mar 29, 2024 3:13 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 82 posts ]  Go to page Previous  1, 2, 3, 4, 5  Next
Author Message
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #61 Posted: Fri Dec 08, 2017 6:36 am 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
moha wrote:
One strong point of this network + search approach seems that it's hard to imagine a game now, where this wouldn't dominate humans.
Imagine a game that does not terminate in a number of moves shallow enough that search to completion is practical (Monte-Carlo requires search to completion in the absence of a reliable evaluation function of non-terminal positions).

A game like football, or gambling on the stock market, for example.

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (not the same as AlphaGo Zero
Post #62 Posted: Fri Dec 08, 2017 8:05 am 
Dies in gote

Posts: 37
Location: Frankfurt, Germany
Liked others: 2
Was liked: 19
Rank: IGS 9k EGF 6k
KGS: Revilo
IGS: Revilo
moha wrote:
Revilo wrote:
I've been out of the trade for a while but I'd say that the specs do not seem too shabby.
Where were they actually described? I only recall seeing things like "1G hash, 64 threads" - but 64 at which kind of hardware? (OC, we can roughly guess from the pos/s given.)


Honestly, I've made the exact educated guess you suggest. I've also assumed that the paper doesn't give evals per seconds but nodes per second - because Stockfish doesn't report its eval calls.

On my Galaxy S8, Stockfish is running at about 3000k nodes per second. We can guestimate from here that they deployed Stockfish on a nice little cluster to achieve 70000k. We can also try to guestimate from the other direction by comparing against the stats given by Stockfish's test cluster: http://tests.stockfishchess.org/tests. It is stated that 459 cores on 99 machines results in roughly 740.000k nodes per second - about 10 times more than what the setup Alpha played against reached. So let's say, 10 quads or something like that, roughly. I hope they are going to be more specific in the announced detailed paper.

Anyway, I judge this to be a quite ok matchup. Stockfish has a fast but rather dumb evaluation function, so it needs raw NPS to go deep into the search space. Alpha has a slow but sophisticated eval (neural network) so it should have proportionally less NPS because otherwise it would be an uneven contest.

Stockfish developer Marco Costalba has actually acknowlegded that this is a huge achievement:

Quote:
I have read the paper: result is impressive!

Honestly I didn't think it was possible because my understanding was that chess is more "computer friendly" than Go....I was wrong.

It is true, SF is not meant to play at its best without a book and especially 1 fixed minute per move cuts out the whole time management, it would be more natural to play with tournament conditions, but nevertheless I think these are secondary aspects, what has been accomplished is huge.


(see here: http://talkchess.com/forum/viewtopic.php?p=741307#741307)

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #63 Posted: Fri Dec 08, 2017 9:17 am 
Lives in sente

Posts: 727
Liked others: 44
Was liked: 218
GD Posts: 10
Criticism from Computer Shogi community (in English) http://www.uuunuuun.com/single-post/201 ... ogi-engine

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #64 Posted: Fri Dec 08, 2017 11:21 am 
Lives in sente

Posts: 727
Liked others: 44
Was liked: 218
GD Posts: 10
DeepMind is the talk of the town at NIPS 2017 conference

David Silver presents AlphaZero. He didn't try AlphaZero vs. Strongest version of AlphaGo Zero yet, and it might take at least one year 'if' DeepMind decide to go opensource.
Image
And this slide is basically a declaration of war to handcraft system
Image

And DeepMind take full feedback/criticism from at least two presenters I found
Image
Image

In which Demis Hassabis defend his AI approach himself (video)
https://twitter.com/thinkmariya/status/ ... 3281185793

The last one, Oriol Vinyals presents DeepMind's progress on Star Craft AI (in which I wish them the best of luck), this is the slide comparison to Go
Image

I think the paper itself is hotter (and controversial) than even AlphaGo in Nature paper (excluding Lee Sedol match hype) because it tackles computer chess, computer shogi, and computer go community at once. So the feedback/reaction is very strong.


This post by pookpooi was liked by: jptavan
Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #65 Posted: Fri Dec 08, 2017 12:15 pm 
Lives in sente

Posts: 902
Location: Fort Collins, CO
Liked others: 319
Was liked: 287
Rank: AGA 3k
Universal go server handle: jeromie
I don’t think the response is necessarily more heated because of the involvement of the chess and shogi communities; it’s more heated because they claim to have developed a generalizable AI algorithm. Since their examples thus far have used perfect information games, there is a question of how applicable this approach is to other types of problems. Naturally, Deepmind is more optimistic than some other researchers. :-)

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #66 Posted: Fri Dec 08, 2017 8:09 pm 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
jeromie wrote:
how applicable this approach is to other types of problems
a question beyond the narrow focus of a Go forum, but a key question for AI and its investors. The short answer is that dcnn has wideranging applicability, as it is a generic pattern discrimination technique (so, for example, one day dentists will use it to tell them whether a tooth needs repair); the long answer is that the scope of future enhancements to dcnn cannot be circumscribed because we do not yet know what those future enhancements will be - already there are people working on expanding dcnn functionality to tasks such as natural language and edge detection.

i imagine that one future development will be the synthesis of composite networks of dcnns that can embrace hierarchical conceptions, and another the development of learning techniques less regimented and more focussed than broadscale hill climbing through static arrays.

_________________
i shrink, therefore i swarm

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #67 Posted: Sat Dec 09, 2017 3:19 am 
Oza

Posts: 3647
Liked others: 20
Was liked: 4626
Here is what most of the world's top chess players think, including how they cope in a world already dominated by chess computers. Lessons for go pros?

https://www.chess.com/news/view/alphaze ... ish-author

From this and other articles it seems that chess and shogi players, while admitting the impressiveness of the achievement, are dissatisfied at the conditions chosen for their game's representative. DeepMind may have blundered a little there and unnecessarily taken some shine off their achievement. Artificial stupidity still rules, OK?

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #68 Posted: Sat Dec 09, 2017 6:32 am 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
John Fairbairn wrote:
Artificial stupidity still rules, OK?

Nice demonstration from FineArt, which can beat up top pros but filled in its own territory instead of passing so lost to a weaker bot in the AI Ryusei: forum/viewtopic.php?p=225812#p225812

(and I agree about the match conditions taking the shine off).

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #69 Posted: Sat Dec 09, 2017 12:15 pm 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
HermanHiddema wrote:
If you post content, people should be free to criticize that content.
Below is my recent output. Feel free to criticise its content.

The Hierarchy of the Imagination
Conceptual structures are intrinsically hierarchical, regardless of whether the reality conceived is itself hierarchical. Examples are given in two different domains.

Learning to SWIM
Mechanisms are described by which a model of conceptual reasoning about Go can learn new techniques from its own analyses of expert moves and assimilate expert advice.

The second one contains an example of a conceptual reasoning machine learning from Alphago zero and from Michael Redmond.

_________________
i shrink, therefore i swarm

Top
 Profile  
 
Offline
 Post subject:
Post #70 Posted: Sun Dec 10, 2017 12:29 am 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
Tech review

Top
 Profile  
 
Offline
 Post subject: Re:
Post #71 Posted: Sun Dec 10, 2017 5:30 am 
Lives in gote

Posts: 311
Liked others: 0
Was liked: 45
Rank: 2d
EdLee wrote:
So it seems AZ excels in the kitchen as well. :)

For the earlier discussion about intelligence, and whether these RL successes may count as such: I wrote that intelligence is ability to solve previously unseen problems, almost the opposite of what reinforcement learning does. This felt a bit weak argument then, but I now convinced myself further with this analogy.

Consider animal behaviour, instincts in particular. Animals can solve complex problems but fail at simple ones if only minor things change and the behaviour suggested by instict is not working anymore. IMO evolutional selection, genetics and mutation = reinforcement learning, where the reward function is survival and reproduction. And instinct vs intelligence = animal vs human behaviour = RL vs intellect. So:

1. random play = eons of failures before success, then eons of failures again
2. reinforcement learning = eons of failures before success, then repeating success (until the problem changes)
3. intelligence = success reasonably soon, using knowledge from other domains

Top
 Profile  
 
Offline
 Post subject: Re: Re:
Post #72 Posted: Sun Dec 10, 2017 6:52 am 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
moha wrote:
instinct vs intelligence = animal vs human behaviour = RL vs intellect
Animals, plants and bacteria learn to improve the efficacy of their behaviour too. See for example pages 90 and 126 of http://sites.google.com/site/djhbrown2/LC1to5.doc

Top
 Profile  
 
Offline
 Post subject:
Post #73 Posted: Sun Dec 10, 2017 4:55 pm 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
The Economist: $ in AI.

Top
 Profile  
 
Offline
 Post subject: Re:
Post #74 Posted: Mon Dec 11, 2017 12:06 am 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
The Economist: $ in AI wrote:
...The most important is whether AI will always depend on vast amounts of data... A competing vision of AI stresses simulations, in which machines teach themselves using synthetic data or in virtual environments. Early versions of a program developed to play Go, an Asian board game, by DeepMind...
Trust the doublethinkThe Economist to mangle a couple of facts to spin a thinly-disguised corporate advertisement in the vein of "Persil Washes Whiter" to suck in more investors by claiming that Alfie Baby doesn't rely upon big data [so she must be really intelligent and will take over the outside real world without needing any real-world data as well as triumphing over the inside world of a board game of limited depth].

Propaganda Works - You Know This
So how about a little propaganda for understanding?

Top
 Profile  
 
Offline
 Post subject: Re:
Post #75 Posted: Mon Dec 11, 2017 1:11 am 
Lives with ko

Posts: 129
Liked others: 20
Was liked: 17
EdLee wrote:
Quote:
Nobody has a self-teaching chess program that can fight with Houdini or Komodo. That’s a fantasy.
Nice if someone finds the clip when Letterman asked Kasparov, (paraphrasing) "Do you think the chess computer will ever beat the best humans?" and Kasparov said no. (This might be shortly after he'd beaten Deep Thought. ...Early 1990's?)

Famous last words, indeed.

I'm late to the party, just recognized AlphaZero --- I was not expecting that. At least I was surprised, Google continued their aggressive nuclear attacks on everything they could find ;)
O-M-F-G

Top
 Profile  
 
Offline
 Post subject: Re: Re:
Post #76 Posted: Mon Dec 11, 2017 7:19 am 
Lives in sente

Posts: 1037
Liked others: 0
Was liked: 180
djhbrown wrote:
The Economist: $ in AI wrote:
...The most important is whether AI will always depend on vast amounts of data... A competing vision of AI stresses simulations, in which machines teach themselves using synthetic data or in virtual environments........... [so she must be really intelligent and will take over the outside real world without needing any real-world data as well as triumphing over the inside world of a board game of limited depth].

Propaganda Works - You Know This


Except I think this is an example of economists not understanding other areas of science and mathematics. A misunderstanding thinking that "learn from zero" means having learned from NO data from the "real" world (in this case the limited definition of the game go). Other information about the real world, vast amounts of information, is indeed irrelevant to learning how to solve THIS problem << learn to play go very well >>

Yes, it turns out that the set of information needed (by alpaha go zero) turns out to be small, but in problems like this, discovering the minimalist set, the necessary and sufficient set, is one of the difficulties. The reason I said "blindness" on the part of the economists is that they don't relate this to THEIR problem. They imagine that they NEED vast amounts of information about the real world (to make their predictions) without seeing that the reason for that is that they don't know which bits of information from the real world are relevant and which are not. And that appearances can be deceptive.

Thus before being shown, we would have thought that alpha go zero would have needed more information << information we cannot see is redundant to the minimal set of information that is sufficient >>

They also do not see what the (generalized) process might be able to do. Given a problem and a PROPOSED minimalist set of information, can the neural net learn to solve the problem "from zero" --- with "from zero" meaning "from just THIS minimalist set of information". In which case that set of information IS a sufficient set.

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #77 Posted: Mon Dec 11, 2017 8:34 am 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
Mike Novack wrote:
Except I think this is an example of economists not understanding other areas of science and mathematics.

The Economist is a general newspaper (in magazine format), it's not actually written solely or indeed much by economists. (Or were you joking?)

Top
 Profile  
 
Offline
 Post subject:
Post #78 Posted: Mon Dec 11, 2017 8:42 pm 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
Quartz: AI complexity ( sorry about the ads )
Quote:
In school we ask students to put it in their own words to prove that they’ve understood something, to show their work, justify their conclusion. <snip> And now we’re expecting that machines will be able to do the same thing.
(emphasis added)
When entity A (eg. student) tries to explain something to entity B (eg. teacher; or vice versa), the above makes an implicit but vital assumption that an explanation exists that B can digest.

This assumption may be true if A's understanding and computational power are roughly similar to B's, or even if A's (slightly) > B's.

But if A's >> B's, it's not so clear why it must be true.

Samples:
A: Einstein (at the publication of his relativity papers ); B: top physicists, general public (at that time; even today).
A: Feynman (on quantum mechanics); B: his dad, general public.
A: average child; B: cat.
A: AlphaZero; B: top pros.

Top
 Profile  
 
Offline
 Post subject: Re: AlphaZero paper discussion (Mastering Go, Chess, and Sho
Post #79 Posted: Mon Dec 11, 2017 9:41 pm 
Lives with ko

Posts: 129
Liked others: 20
Was liked: 17
pookpooi wrote:
Criticism from Computer Shogi community (in English) http://www.uuunuuun.com/single-post/201 ... ogi-engine

Concerns, yes, but not surprising, Deepmind probably did not find decent information on what is state of the art in Computer Shogi. As one cannot find decent software simply. You fight your way through Google translated japanese pages and end up finding GUIs appealing as software from the 90s. E.g., in Shogi, best I found the other day was ShogiGUI, which, by chess software standards, was still clunky retro style software. And you find like 3 people commenting on it on the internet. And first you do is tell it to install in english - and still get a japanese UI. So first thing to do is find the option o change language. Which is not an easy task.
Or Shogidokoro. Oh well, also like software from 1990. IMO Shogi software development is lightyears behind western chess. Go only lightmonths ;)

Top
 Profile  
 
Offline
 Post subject: Re:
Post #80 Posted: Tue Dec 12, 2017 1:29 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
EdLee wrote:
Quartz: AI complexity ( sorry about the ads )
Quote:
In school we ask students to put it in their own words to prove that they’ve understood something, to show their work, justify their conclusion. <snip> And now we’re expecting that machines will be able to do the same thing.
(emphasis added)


I'm sorry, Dave.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 82 posts ]  Go to page Previous  1, 2, 3, 4, 5  Next

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group