ICEINSPACE
Most Read Articles
Moon Phase
CURRENT MOON Waxing Gibbous
84.2%
The Sun Now
Time Zones
Sydney*
3:39 pm
Perth
12:39 pm
Auckland*
5:39 pm
New York
11:39 pm
Paris
5:39 am
GMT
4:39 am




Go Back   IceInSpace > General Astronomy > Astronomy and Amateur Science

Reply
 
Thread Tools Rate Thread
  #1  
Old 07-12-2018, 09:39 AM
gary
Registered User

gary is offline
 
Join Date: Apr 2005
Location: Mt. Kuring-Gai
Posts: 4,891
How to become a master at chess and just about any other game

In an article at the IEEE Spectrum web site, Philip E. Ross reports
on a paper published today in Science by David Silver et. al. of
DeepMind in London on their software program called AlphaZero.

As reported earlier, AlphaZero trained itself, without human intervention &
in a matter of hours, how to to play Go, chess & shogi sufficiently well to be
able to beat other champion programs of those games.

Quote:
Originally Posted by Philip E. Ross, IEEE Spectrum
Deep Blue was a monster of a machine built solely to play chess, and its 1997 victory over Kasparov was not overwhelming. Today, though, even a smartphone can outplay Magnus Carlsen, the reigning world champion, and do so again and again:

But that smartphone is just a piker compared to the top conventionally programmed chess program, Stockfish. And Stockfish, in turn, is a piker next to AlphaZero, which crushed it after a mere 24 hours of self-training.
Now the DeepMind Researchers claim that the generalisations behind
the neural network algorithms in AlphaZero should make it capable of
mastering a wider variety of games, including multiplayer games.

Quote:
Originally Posted by Philip E. Ross, IEEE Spectrum
“This work has, in effect, closed a multi-decade chapter in AI research,” writes Campbell, who was a member of the team that designed IBM’s Deep Blue, which in 1997 defeated Garry Kasparov, then the world chess champion. “AI researchers need to look to a new generation of games to provide the next set of challenges.”

AlphaZero can crack any game that provides all the information that’s relevant to decision-making; the new generation of games to which Campbell alludes do not. Poker furnishes a good example of such games of “imperfect” information: Players can hold their cards close to their chests. Other examples include many multiplayer games, such as StarCraft II, Dota, and Minecraft. But they may not pose a worthy challenge for long.

“Those multiplayer games are harder than Go, but not that much higher,” Campbell tells IEEE Spectrum. “A group has already beaten the best players at Dota 2, though it was a restricted version of the game; Starcraft may be a little harder. I think both games are within 2 to 3 years of solution.”
Article here :-
https://spectrum.ieee.org/tech-talk/...ntelligence/mb

Science paper "A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play" by Silver et. al. here :-
http://science.sciencemag.org/content/362/6419/1140

Background paper published at the same time in Science, "Mastering board games" by Murray Campbell :-
http://science.sciencemag.org/content/362/6419/1118

Background paper in Science published today, "Chess, a Drosophila of reasoning" by Garry Kasparov :- http://science.sciencemag.org/content/362/6419/1087

Quote:
Originally Posted by Garry Kasparov, Science
I admit that I was pleased to see that AlphaZero had a dynamic, open style like my own. The conventional wisdom was that machines would approach perfection with endless dry maneuvering, usually leading to drawn games. But in my observation, AlphaZero prioritizes piece activity over material, preferring positions that to my eye looked risky and aggressive. Programs usually reflect priorities and prejudices of programmers, but because AlphaZero programs itself, I would say that its style reflects the truth. This superior understanding allowed it to outclass the world's top traditional program despite calculating far fewer positions per second. It's the embodiment of the cliché, “work smarter, not harder.”

AlphaZero shows us that machines can be the experts, not merely expert tools. Explainability is still an issue—it's not going to put chess coaches out of business just yet. But the knowledge it generates is information we can all learn from. Alpha-Zero is surpassing us in a profound and useful way, a model that may be duplicated on any other task or field where virtual knowledge can be generated.
Reply With Quote
  #2  
Old 07-12-2018, 11:23 AM
multiweb's Avatar
multiweb (Marc)
ze frogginator

multiweb is offline
 
Join Date: Oct 2007
Location: Hinchinbrook
Posts: 17,151
AlphaZero... sounds like a military project. Do I smell applications behind game theory?
Reply With Quote
  #3  
Old 07-12-2018, 12:32 PM
gary
Registered User

gary is offline
 
Join Date: Apr 2005
Location: Mt. Kuring-Gai
Posts: 4,891
Quote:
Originally Posted by multiweb View Post
AlphaZero... sounds like a military project. Do I smell applications behind game theory?
Hi Marc,

The company behind it, DeepMind, is a subsidiary of Google.

As you may be aware, in game theory, chess is not a game.
That's because, like noughts and crosses, there really exists out there
a well ordered set of moves that should always result in a win or draw.

The seminal 1944 book, "Theory of Games and Economic Behavior"
by John von Neumann and Oskar Morgenstern, is filled with heavy
mathematical equations but there is a chapter entitled "Poker and Bluffing".

In game theory, poker is defined as a "game" but chess is not.

Nevertheless, the genius of Von Neumann was to prove that even for
"true" games like poker there exists best strategies.

During the late Cold War in the 1940's and 50's, much of Von Neumann's
life was arguably frittered away consulting to US government think tanks
such as the RAND Corporation on the application of game theory to
strategic nuclear war planning.

These days, game theory is more famously applied to predicting the stock
market.

It will be interesting to watch how AlphaZero evolves to handle games of the game theory
type that have incomplete information and where opponents use tactics such as bluffing.

Google will probably want to use it to predict what you want to buy even before you have thought about it.
Reply With Quote
  #4  
Old 07-12-2018, 01:39 PM
AstralTraveller's Avatar
AstralTraveller (David)
Registered User

AstralTraveller is offline
 
Join Date: Mar 2008
Location: Wollongong
Posts: 3,483
Quote:
Originally Posted by gary View Post
Originally Posted by Philip E. Ross, IEEE Spectrum
Deep Blue was a monster of a machine built solely to play chess, and its 1997 victory over Kasparov was not overwhelming. Today, though, even a smartphone can outplay Magnus Carlsen, the reigning world champion, and do so again and again:

But that smartphone is just a piker compared to the top conventionally programmed chess program, Stockfish. And Stockfish, in turn, is a piker next to AlphaZero, which crushed it after a mere 24 hours of self-training.

And here is AlphaZero crushing Stockfish.


https://www.youtube.com/watch?v=pcdpgn9OINs


Follow links on the site to see other AlphaZero games.


Obviously there is much more to developing this AI than just winning a game, noble as that is. However it is also affecting chess. For one thing all important games are now analyised in real time by a chess engine which evaluates who is in the better position (and, I presume, suggests the next move). Clearly the players should be unaware of all this but of course the motivation to get a little bit of assistance is pretty great. There have already been (farcical) accusations along these lines but real cases will surely occur. I also wonder about classical chess where the day's play finishes after 40 moves. What is to stop players (and seconds) from spending he night analyzing the position on a computer? Players have always had help overnight but having AlphaZero takes this to a new level.


It also changes, for the better or worse depending on your perspective, the audience's experience. On one hand the duffer players now know who is in the best position but on the other they no longer have to think hard about who they believe is in the better position.



Finally, it is affecting how the game is played. For one thing, it must improve opening theory. However it may not improve games. The presenter on the site I referenced above feels that players are playing very strong, machine-like moves for about 12-15 moves and then, once they have passed the moves they have memorised, the standard drops. It may not be because the players are any weaker than in the past so much as they more or less have to turn on their brain part way through the match, rather than being deeply involved from the start. Sort of like picking up a job that someone else has started.


Anyway, the genie is out of the bottle and we'll just have to see what eventuates.
Reply With Quote
  #5  
Old 07-12-2018, 04:38 PM
gary
Registered User

gary is offline
 
Join Date: Apr 2005
Location: Mt. Kuring-Gai
Posts: 4,891
Cool

Quote:
Originally Posted by AstralTraveller View Post

However it is also affecting chess.

Finally, it is affecting how the game is played. For one thing, it must improve opening theory.
Hi David,

Thanks for the post, the link and the interesting background as to
how these chess playing programs are affecting chess tournaments.

In the April 1975 edition of Scientific American, Martin Gardner in the
"Mathematical Games and Puzzles" section authored an April Fool's
spoof entitled "Six sensational discoveries that somehow or another
have escaped public attention".

One of these gems was the announcement that there was now a computer
proof that if white opened with "pawn to king's rook 4" that it would
result in a win for white "with a high degree of probability."

Despite the fact the same article included a claim that Leonardo da Vinci
had invented the valve flush toilet, complete with a drawing in the style
of Leonardo of a renaissance man sitting on the can, the fact that it
was an April Fools joke was lost to many readers and some chess
aficionados were aghast that their centuries-old game was now
hardly worth playing.

And the real salt in the wound was that, as I understand it,
"pawn to king's rook 4" had never been regarded as a particularly strong
opening move.

Nevertheless, the prospect of such as spectacular revelation one day being
announced has always stuck in my mind.

Just as when we are young, most of us quickly learn that "if I start here"
in noughts and crosses, "nothing you can do will not result in a win
or at least a draw for me", I've wondered at the prospect that one day,
even though it is exceedingly unlikely, that a deep computer analysis
dropped a bombshell much akin to Martin Garner's April Fools joke,
that is, "if I open with this particular move" and then continue with a
precisely computable set of counter moves, I will always win.

It certainly would break the hearts of a lot of people.

Chess would be regulated to this "simple game" like noughts and crosses
suitable only for young children. Tournaments would cease. Chess board
and chess book sales would all but halt.

Grand masters would hide behind dark glasses, move to Vegas and
turn their hands to Texas hold'em instead!
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 02:39 PM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
Advertisement
Atik Horizon
Advertisement
OzScopes Authorised Dealer
Advertisement
SkyWatcher WiFi Adaptor
Advertisement
NexDome Observatories
Advertisement
Bintel
Advertisement
Lunatico Astronomical
Advertisement
SkyWatcher 2018 Catalogue
Advertisement
Interest Free Finance
Advertisement
FLI Cameras and Imaging Accessories
Advertisement
Astronomy and Electronics Centre
Advertisement