It is currently Sat Nov 18, 2017 9:38 pm

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 39 posts ]  Go to page Previous  1, 2
Author Message
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #21 Posted: Fri Apr 07, 2017 8:21 am 
Dies in gote

Posts: 26
Liked others: 0
Was liked: 5
Rank: Europe 5 dan
KGS: Flashgoe
pookpooi wrote:
Bohdan, the free DCNN program Leela has option to let you play with 25x25 board. Can you tell us how much weaker it is compare to 19x19. You'll have no problem winning in both board size in even game since it has serious bug that prevent it to go beyond 6 dan KGS so you've to give it handicap stones.


Sure thing. I'll play it and upload games here. Nice challenge :)

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #22 Posted: Fri Apr 07, 2017 8:35 am 
Lives in gote

Posts: 597
Liked others: 41
Was liked: 190
GD Posts: 10
Bohdan wrote:
Let us look at the different angle. The Crazy Stone bot without NN palys around 5 dan on an average laptop. It was built by only one man who was doing it mostly for fun having a lot of other stuff to do. Imagine how whole team of very skilled programers and scientists from DeepMind (read Google) could improve an algorithm from a Crazy Stone bot and multiply it by 500 CPU and 200 GPU power. I would never be surprised if optimized version of MC trees can easily beat an average pro.

This remind me of Mogo, French MCTS bot just like Crazy Stone. It famous around 2007-2011 when it challenge pro/top amateur players with super computer, like 512 cores of the Louhi Cray XT4/XT5 supercomputer (and still lose to Catalin Taranu 5p in 7 stones handicap), 200 node supercomputer with 4 cores per node (and lose to Myungwan Kim 8p, also in 7 stones handicap), 640 cores of the Huygens supercomputer in Amsterdam (and lose to Chun-Hsun Chou 9p and Shih Chin 2p in 7 stones handicap game)

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #23 Posted: Fri Apr 07, 2017 8:52 am 
Dies in gote

Posts: 26
Liked others: 0
Was liked: 5
Rank: Europe 5 dan
KGS: Flashgoe
pookpooi wrote:
Bohdan wrote:
Let us look at the different angle. The Crazy Stone bot without NN palys around 5 dan on an average laptop. It was built by only one man who was doing it mostly for fun having a lot of other stuff to do. Imagine how whole team of very skilled programers and scientists from DeepMind (read Google) could improve an algorithm from a Crazy Stone bot and multiply it by 500 CPU and 200 GPU power. I would never be surprised if optimized version of MC trees can easily beat an average pro.

This remind me of Mogo, French MCTS bot just like Crazy Stone. It famous around 2007-2011 when it challenge pro/top amateur players with super computer, like 512 cores of the Louhi Cray XT4/XT5 supercomputer (and still lose to Catalin Taranu 5p in 7 stones handicap), 200 node supercomputer with 4 cores per node (and lose to Myungwan Kim 8p, also in 7 stones handicap), 640 cores of the Huygens supercomputer in Amsterdam (and lose to Chun-Hsun Chou 9p and Shih Chin 2p in 7 stones handicap game)


Pardon me but it is nowhere near Crazy Stone. Mogo strength is kyu level.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #24 Posted: Fri Apr 07, 2017 8:58 am 
Lives in gote

Posts: 597
Liked others: 41
Was liked: 190
GD Posts: 10
Bohdan wrote:
Pardon me but it is nowhere near Crazy Stone. Mogo strength is kyu level.

At that time Crazy Stone is also Kyu level.
At 2007 Amsterdam Computer Olympiad, in 19x19 category, Mogo got Gold medal while Crazy Stone got Silver medal.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #25 Posted: Fri Apr 07, 2017 9:03 am 
Dies in gote

Posts: 26
Liked others: 0
Was liked: 5
Rank: Europe 5 dan
KGS: Flashgoe
pookpooi wrote:
Bohdan, the free DCNN program Leela has option to let you play with 25x25 board. Can you tell us how much weaker it is compare to 19x19. You'll have no problem winning in both board size in even game since it has serious bug that prevent it to go beyond 6 dan KGS so you've to give it handicap stones.


You mean it fails to see a snapback?

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #26 Posted: Fri Apr 07, 2017 9:04 am 
Dies in gote

Posts: 26
Liked others: 0
Was liked: 5
Rank: Europe 5 dan
KGS: Flashgoe
pookpooi wrote:
Bohdan wrote:
Pardon me but it is nowhere near Crazy Stone. Mogo strength is kyu level.

At that time Crazy Stone is also Kyu level.
At 2007 Amsterdam Computer Olympiad, in 19x19 category, Mogo got Gold medal while Crazy Stone got Silver medal.


I see. Anyway I am sure that you can reach a pro level by only using a MC method.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #27 Posted: Fri Apr 07, 2017 9:17 am 
Tengen

Posts: 4230
Location: Cambridge, UK
Liked others: 158
Was liked: 2082
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
Bohdan wrote:
Anyway I am sure that you can reach a pro level by only using a MC method.

I agree, if Zen kept on improving like it was before AlphaGo made neural nets all the rage (not that is was the first to use a policy network, only value network) then I would be fairly sure it would beat a low pro at most 5 years from back then (it was about European 5 to 6 dan in 2015: FJ Dickhut German 6d beat it 3-1) and quite possible sooner, and faster if it got more developers/processing power. But to go from that to beating Lee Sedol is quite a jump that I thought would take 10 years or more without the value network revolution.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #28 Posted: Fri Apr 07, 2017 9:31 am 
Oza

Posts: 2182
Location: ʍoquıɐɹ ǝɥʇ ɹǝʌo 'ǝɹǝɥʍǝɯos
Liked others: 237
Was liked: 659
Rank: AGA 5d
GD Posts: 4312
Online playing schedule: Every tenth February 29th from 20:00-20:01 (if time permits)
The original poster seems to think that power consumption should be comparable for an equal match. I immediately thought of it a different way and read the entire thread look for a mention of it. No one else did, so here it is. What do you think?

Trying to compare robots with humans is like comparing apples to oranges. I was reminded of the character Data in the old Star Trek series. He was a humanoid robot, self contained although able to access some universal database. This is ultimately where computers are heading. So to me, instead of comparing power usage, how about comparing the human to a computer which is totally self contain and able to fit into a humanoid body? Could the computer sit down at a table opposite a human and, with no external power input, win the game. An external humanoid appearance would be window dressing - it is the size which counts.

Just an alternative perspective and one which many will take issue with.

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #29 Posted: Sat Apr 08, 2017 7:58 am 
Lives in gote

Posts: 436
Liked others: 1
Was liked: 38
Rank: KGS 5 kyu
DrStraw wrote:
The original poster seems to think that power consumption should be comparable for an equal match. I immediately thought of it a different way and read the entire thread look for a mention of it. No one else did, so here it is. What do you think?

Trying to compare robots with humans is like comparing apples to oranges. I was reminded of the character Data in the old Star Trek series. He was a humanoid robot, self contained although able to access some universal database. This is ultimately where computers are heading. So to me, instead of comparing power usage, how about comparing the human to a computer which is totally self contain and able to fit into a humanoid body? Could the computer sit down at a table opposite a human and, with no external power input, win the game. An external humanoid appearance would be window dressing - it is the size which counts.

Just an alternative perspective and one which many will take issue with.


Can we use batteries or is that considered an external source of energy? By that logic human player should be starved for days since our external energy source is food.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #30 Posted: Sun Apr 09, 2017 2:24 pm 
Oza

Posts: 2182
Location: ʍoquıɐɹ ǝɥʇ ɹǝʌo 'ǝɹǝɥʍǝɯos
Liked others: 237
Was liked: 659
Rank: AGA 5d
GD Posts: 4312
Online playing schedule: Every tenth February 29th from 20:00-20:01 (if time permits)
Krama wrote:
DrStraw wrote:
The original poster seems to think that power consumption should be comparable for an equal match. I immediately thought of it a different way and read the entire thread look for a mention of it. No one else did, so here it is. What do you think?

Trying to compare robots with humans is like comparing apples to oranges. I was reminded of the character Data in the old Star Trek series. He was a humanoid robot, self contained although able to access some universal database. This is ultimately where computers are heading. So to me, instead of comparing power usage, how about comparing the human to a computer which is totally self contain and able to fit into a humanoid body? Could the computer sit down at a table opposite a human and, with no external power input, win the game. An external humanoid appearance would be window dressing - it is the size which counts.

Just an alternative perspective and one which many will take issue with.


Can we use batteries or is that considered an external source of energy? By that logic human player should be starved for days since our external energy source is food.


You clearly did not watch Star Trek. :)

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #31 Posted: Mon Apr 17, 2017 1:26 pm 
Lives in gote

Posts: 436
Liked others: 1
Was liked: 38
Rank: KGS 5 kyu
DrStraw wrote:
Krama wrote:
DrStraw wrote:
The original poster seems to think that power consumption should be comparable for an equal match. I immediately thought of it a different way and read the entire thread look for a mention of it. No one else did, so here it is. What do you think?

Trying to compare robots with humans is like comparing apples to oranges. I was reminded of the character Data in the old Star Trek series. He was a humanoid robot, self contained although able to access some universal database. This is ultimately where computers are heading. So to me, instead of comparing power usage, how about comparing the human to a computer which is totally self contain and able to fit into a humanoid body? Could the computer sit down at a table opposite a human and, with no external power input, win the game. An external humanoid appearance would be window dressing - it is the size which counts.

Just an alternative perspective and one which many will take issue with.


Can we use batteries or is that considered an external source of energy? By that logic human player should be starved for days since our external energy source is food.


You clearly did not watch Star Trek. :)


I think we could fit new tesla battery inside a human body (total volume not the shape). plus some volume for the computer itself (mobo + gpu). So I guess we could create a robot that is self powered and could move a limb in order to play a move. The battery would probably last for few hours.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #32 Posted: Fri Apr 21, 2017 1:53 am 
Beginner

Posts: 7
Liked others: 0
Was liked: 1
Rank: 8k
A computer that is self contained and uses the same amount of energy as a human will certainly sooner rather than later beat humans at go, probably less than one human lifetime from now. I base that on the advances that have been made in computer software and hardware during my lifetime. There is no reason to expect that this rate of advance will slow down. I would actually be very surprised if a device that is the size of a modern smartphone and uses the same amount of power could easily beat a pro 20 or 30 years from now.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #33 Posted: Fri Apr 21, 2017 12:39 pm 
Lives in sente

Posts: 857
Liked others: 0
Was liked: 139
Peter1 wrote:
A computer that is self contained and uses the same amount of energy as a human will certainly sooner rather than later beat humans at go, probably less than one human lifetime from now. I base that on the advances that have been made in computer software and hardware during my lifetime. There is no reason to expect that this rate of advance will slow down.


Computers barely exist within a lifetime so far.

HOWEVER -- I just thought of another way to look at the original question that might be more meaningful to folks. A device using that amount of energy over the length of time of a go game might be a PORTABLE device.

So --- how soon do we expect that level of play on something like a portable workstation class device? << resembles your laptop in size and weight and can run a couple hours on its battery -- but these devices have more powerful CPUs and GPUs than your typical desktop >> BTW, THAT is the "advance" question. A great deal of the research/engineering goes into creating components not more powerful but can have the same or close to the same using a less power precisely so can be in a portable device with useful operating time not plugged in.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #34 Posted: Sun Apr 23, 2017 8:36 am 
Tengen

Posts: 4042
Location: North Carolina
Liked others: 411
Was liked: 662
Rank: AGA 3k
GD Posts: 65
OGS: Hyperpape 4k
What's the current state of the art? Crazy Stone is dan level on smartphones now. I think Leela 0.9.0 with GPU acceleration is borderline for finishing a whole game on my laptop (i7 7500U ultrabook). It just now burned 16% of my battery playing 25 moves for each side (computer and me), but I don't know quite how strong it is on this machine.

Edit: forgot I had an integrated GPU which both will save power and hurt performance. In fact, Leela's author says this GPU may perform worse than not using it...I really did not care about GPUs when I bought this machine.

_________________
Occupy Babel!

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #35 Posted: Sun Apr 23, 2017 9:47 am 
Dies with sente

Posts: 101
Liked others: 21
Was liked: 16
hydrogenpi7 wrote:
AlphaGo used thousands of CPU's and hundreds of GPU's to inch out Lee Sedol. The human brain only uses about 20 to 25 Watts.


Great idea, energy as the big equalizer.
I would add: humans should only be allowed the same amount of WC breaks as machines need: "WC break time parity"


This post by alphaville was liked by: Waylon
Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #36 Posted: Tue Apr 25, 2017 2:11 pm 
Lives in gote

Posts: 436
Liked others: 1
Was liked: 38
Rank: KGS 5 kyu
alphaville wrote:
hydrogenpi7 wrote:
AlphaGo used thousands of CPU's and hundreds of GPU's to inch out Lee Sedol. The human brain only uses about 20 to 25 Watts.


Great idea, energy as the big equalizer.
I would add: humans should only be allowed the same amount of WC breaks as machines need: "WC break time parity"


But human brain on it's own can't do anything. Brain requires functioning body in order to function itself.

Average male needs 2500 calories per day. So if match lasts around 5 hours I guess we could just divide 2500 by 5 for simplicity?

500 calories is 2092000 Joules.
Power (watts) = Joules per unit of time (seconds).
5 hours = 18000 seconds.
So in those 5 hours we get around 116 watts of power that average human body uses.
Did I miss something in the math?

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #37 Posted: Tue Apr 25, 2017 3:08 pm 
Oza

Posts: 2463
Location: DC
Liked others: 145
Was liked: 435
Universal go server handle: skydyr
Online playing schedule: When my wife is out.
Krama wrote:
alphaville wrote:
hydrogenpi7 wrote:
AlphaGo used thousands of CPU's and hundreds of GPU's to inch out Lee Sedol. The human brain only uses about 20 to 25 Watts.


Great idea, energy as the big equalizer.
I would add: humans should only be allowed the same amount of WC breaks as machines need: "WC break time parity"


But human brain on it's own can't do anything. Brain requires functioning body in order to function itself.

Average male needs 2500 calories per day. So if match lasts around 5 hours I guess we could just divide 2500 by 5 for simplicity?

500 calories is 2092000 Joules.
Power (watts) = Joules per unit of time (seconds).
5 hours = 18000 seconds.
So in those 5 hours we get around 116 watts of power that average human body uses.
Did I miss something in the math?


Well looking at it from the other side, a significant portion of the things the brain is doing have nothing to do with go. Perhaps you need to cut the amount of energy spent to the amount actually used on go, as opposed to things like breathing, digestion, locomotion etc. Or maybe the computer should be required to handle all those functions as well?

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #38 Posted: Sun May 21, 2017 2:53 pm 
Lives in gote

Posts: 436
Liked others: 1
Was liked: 38
Rank: KGS 5 kyu
I think we shouldn't because all parts of the body are quite important for the body and thus the brain to function. Our eyes and vision are needed, hand motions, and bunch of other mental stuff like keeping your spirits up (motivation), fear etc.
Without a healthy body brain can't properly function.
Also a fan that is cooling the hardware has nothing to do with the game of go itself but without it hardware would overheat.

Top
 Profile  
 
Offline
 Post subject: Re: Will AI ever beat humans at Go with energy parity?
Post #39 Posted: Sun May 21, 2017 3:23 pm 
Lives with ko
User avatar

Posts: 245
Liked others: 12
Was liked: 30
Rank: NR
this whole thread should be in Off-topic, because it's clear to any reader other than the writers that their conversation is not about Go per se, but about man vs machine in general. And it's clear to any amateur shrink - it's so obvious, you don't need to be a pro - that the motivation for the conversation is a deep sense of insecurity+hubris, a common dissonance, especially among those schooled in the tradition that they are cast in the image of God, whereas it is clear that Dog is God when it comes to Go - until, that is, when Swim is born and will grab the monkey by the tail and throw it out of the tree.

Instead of fretting about how impotent Ke Jie is when faced by the Dog - or, in my case, even by her little sister / mini-clone Leela - you could instead marvel at how marvellous Aplysia is; how it manages to do so much with so few neurons. Or, indeed, how wise E.Coli is, with none at all!

The computational machinery of Nature is the greatest Wonder of both the Ancient and Modern world - sure, we can't out-bulldoze a bulldozer, or out-climb a thimbleful of petrol up a hill. but our machinery has enabled us (or, rather, them, since not many people have) to figure out that babies are not gifts from the Moon, and that the Sun does not rise in the East and set in the West, and does not die and become reborn every Christmas (or Easter - take your pick) as the Good Bok says, which is wrong about just about everything, but still casts the stone-minds of the gullible.

When it comes to power/weight ratios, i have been mightily impressed by the remarkable progress of hardware technology over the last 50 years, and mightily disappointed by the equally remarkable lack of progress of software technology over the same period. If anything, it seems to be going backwards.

_________________
i ink, therefore i arm

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 39 posts ]  Go to page Previous  1, 2

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: Google [Bot] and 3 guests


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