Slashdot Log In
IBM Working on Brain-Rivaling Computer
Posted by
CmdrTaco
on Tue Nov 19, 2002 10:11 AM
from the will-it-get-bored-too dept.
from the will-it-get-bored-too dept.
Obdurate writes "The first supercomputers to approach and even surpass the processing power
of the human brain are to be built by IBM, under a $184M contract
announced by the US Government yesterday.
ASCI Purple and Blue Gene/L will be the fastest and most powerful machines built,
with a combined capacity equal to the 500 best of todays computers."
This discussion has been archived.
No new comments can be posted.
IBM Working on Brain-Rivaling Computer
|
Log In/Create an Account
| Top
| 627 comments
(Spill at 50!) | Index Only
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
|
2
(1)
|
2
uhu (Score:5, Interesting)
Re:uhu (Score:5, Funny)
Hate to be a dick, but.
It's people like you that make it hard for people like me to get laid. It's not just that you know a bunch of stuff about technology. It's that you know a bunch of stuff about technology that doesn't exist. For example, last week I was forced to break up a conversation that two of my cow-orkers were having on the relative merits and drawbacks of the different types of transporters used by the different races. Had that conversation continued, it might have made it impossible for any geek to EVER GET LAID AGAIN.
Just the hearing phrase "computer scientist" causes most women to stop ovulating immediately already. Let's not make things worse on ourselves.
Re:uhu (Score:5, Funny)
MASPAR (Score:5, Insightful)
To tie in an ever popular
Wait a minute! (Score:5, Interesting)
Hold on there!
Our brains are fine for huge linear calculations. Better than most calculators in fact.
Autistic savants....
Rain Main. That kind of thing.
There was a kid I knew in high school that could find cube roots for eight digit numbers nearly instantly but he couldn't recognize his brother's face in a picture.
My personal theory is this: Human brains are like a computer (about a million orders of mangitude more complex though). Most people have that all tied up in hardware dedicated to things like jobs, girl friends, football etc. etc.
John, my autistic friend in high school, hadn't dedicated the hardware to anything in particular, but he still had it available. He was lacking in a lot of things, but sheer processing power and memory he had in spades.
As a side story, another friend of mine in high school had epilepsy, and it kept getting worse. He eventually had brain surgery where they severed his corpus callosum. After that, he couldn't add single digit numbers if he closed his right eye. If he closed his left, he couldn't recognize faces. Just kind of shows how the brain works as a parallel system.
Re:uhu (Score:5, Funny)
Smarter! was: Re:uhu (Score:4, Interesting)
Re:uhu (Score:5, Funny)
SPEC-brain exists and it's almost what you think.. (Score:4, Interesting)
SPEC brain scans are actually quite commonly usely used to understand brain activity. Here's a study that shows how it's used:
http://neuro-www.mgh.harvard.edu/forum_2/ADHDF/
Re:uhu (Score:5, Informative)
Even so, the human brain is rated somewhere at millions of gigaflops. Quite interesting. Here are some articles (google for some more):
http://www.intel.com/pressroom/archive/speeches/j
http://zinos.com/cool/zinos/scan/se=AR002649/sp=v
Adding numbers (Score:5, Insightful)
Look at it this way. Go outside, on a windy day (adding more variables to the mix) and have someone throw you a football/basketball/baseball/frisbee/whatever. It probably takes 3-4 seconds at most for the ball to reach you, and looooong before that, your brain completed a monstrous calculus problem. It figured in the position of the thrower, the wind velocity and direction, direction/speed of the ball, the ball's arc of travel, and in the next split second, sent signals to your legs and feet to move your body to the ball's expected landing spot.
But wait, it's the ball's landing spot minus about five feet, because your brain figures you want to be positioned to catch the ball when it's about 4-5 feet off the ground. It simultaneously sends signals to your hands and arms, positioning them to catch the ball, taking into account the ball's speed, size and mass.
A lot of calculations in an extremely short period of time! And, if you think that's impressive for a human brain, the brain in that dumb mutt of yours in the back yard can do the same thing when you toss him a tennis ball.
Re:Adding numbers (Score:4, Informative)
What, praytell, can "human intuition" possibly be other than the result of the brain taking information and acting on it? The analogy between a computer and a human brain has all sorts of problems.
Nevertheless, there is no such thing as "human intuition". Brains are made of neurons. Chemical and electrical signaling between the neurons is the only thing that causes anything to happen in our brains. There is no humonculus controlling anything. There is no random number generator. Human intuition may be described, IMHO, as logical extrapolations based on imperfect knowledge. It is not some mystical, non-computational characteristic of neurology.
The entire concept that we can match up a machine's computation to the brain's is trivializing how the brain functions.
The brain is relatively simply at a basic level. Chemicals and electric signals are exchanged by various neurons. This represents the exchange of information, some meaningful, some not, some we just don't know about. Certain regions of the brain are responsible for processing visual data (much of the "conscious" brain could be viewed as a massive extension of the eye).
We break down each function related to the problem and track it to the subsystem, breaking everything down into smaller and smaller and more discrete processes, and it all begins to look very much like simple computational problems. We're used to dealing with digital computers and our analysis of how to solve problems with digital computers is certainly not applicable to the brain on a one-to-one basis -- that is just nuts.
The short reply to your assertion is, however, that the only way we will ever understand the functioning of the "brain" and the rest of the related nervous system is to break it down into little parts, i.e. trivialize it.
I was able to catch a football before I even studied mathematics, let alone arithmetic. There is no calculus problem being solved.
But I'll bet that you didn't learn how to catch a ball without getting stoved fingers, missing a bunch of them, dropping balls on occasion from mis-judging speed, height, the position of your body, etc.
Memory, experience, and the brain's wonderful ability to track moving things (likely a residual survival skill) easily do this without requiring conscious thought on your part.
The fact that you are unaware of the process and the calculations being made does not mean that they are not being made. Are you aware of the temperature calculations for when you bump the stove? ("I wonder how hot this is...hmmm...it feels as though it might cause third degree burns in 1.2 seconds...oh...it has already been 3.2 seconds...I'd better remove my hand.") Much is going on "behind the curtain". Consciousness appears to be related to only a very little of what we do on a regular basis.
guac-foo
Re:Adding numbers (Score:4, Insightful)
Yes there is.
Just because you hadn't been taught how to manipulate manmade concepts such as symbols and numbers and call the process 'calculus' doesn't mean your brain hadn't formed skills to calculate changes in dozens of variables such as position/velocity over time and act on the results.
You might as well say "I never learned biochemistry until college. Prior to that, eating involved no complex carbohydrates being digested because I didn't know how."
Human procesing power article (was: Re:uhu) (Score:5, Informative)
A nice article I found about the subject is http://www.transhumanist.com/volume1/moravec.htm [transhumanist.com]
Nice graph from the document: http://www.transhumanist.com/volume1/All_things_07 5.jpg [transhumanist.com]
Reinout
Re:uhu (Score:5, Informative)
Re:uhu (Score:5, Informative)
There is evidence that the human brain works in a holographic manner. Ref: book [amazon.com].
The basic idea is that all the dentrites for a rooted interference network. The electrical stimuli probagate through the roots and electromagnetically interfere with each other in very complex ways that are unique to each brain's particular wiring.
The particular orientation, size and speed of the wiring isn't important. What is important is that there is sufficient randomness and interconnectedness.
One of the most striking attributes of holography is that when two images are super-imposed apon one another and a representation of their signal is passed through the holographic film-plate, a correlation is saved to the medium. Now if either of the two original signals (representing one of the original physical objects) is passed into the holographic medium, then the second object will be reinvigorated. As en axample. Light Beam splitting and bouncing two object to record to a holographic film plate allows for shutting off one beam and still being able to see both images reflected off the holographic plate. This is [presumably] the property of association in the brain. Thus we associated all things we experience to all other things we experience. The most vivid the experiance, the greater the tendency of that experience to be reflected apon when it's corrolated activities recur. From this, the mere recollection of the original vivid experience is enough to associate it with newer experiences. Thus our unique human defining characterists have to do with the particular order that vivid experiences occur and what things we associate them with.
Further, a hologram stores an infinite amount of information with decreasing precision. If you take a holographic plate and break into into a thousand irregular pieces... Each piece contains the complete information necessary to reproduce the holographic image (and associated correlations), however, the image becomes blurrier and blurrier the smaller the fragment. Mathmatically, this is a convolution operation (If I'm not mistaken), where every piece of data manipulates every other piece of data, so every saved piece of data (atom) is some combination of all data. But the data is useless without some amount of adjacenta data to re-extract some approximation of the original. It's like taking the Fourier transform of a signal, then band-limiting it in some complex way, then converting back to the time-domain (e.g. mp3 compression, but for the entire audio file instead of in blocks). You get a fuzzy view of reality, yet some patterns still remain very apparent.
This is [possibly] the fundamental problem with modeling the human brain in a computer. A computer can not base future decisions on a linear combination of all past experiences, since it would very quickly be overwhelmed with data, it's an O(n!) problem. It would not be much better off if it attempts to simply take the fouier transform of all it's input data and perform statistical analysis. This assumes a single transformation model (such as sinusoidal decomposition). Instead, (if I'm interpreting the book correctly), the holographic model would be an irregular transformation with a highly complex (though analog and fast) decomposition based on probagation delays through random lenghted dentriets all generating complex interference patters. This allows for a large number of orthogonal patterns to be represented in the transformed domain (as opposed to simple sinusoidal patterns with Fourier).
The first thing to note is that analog representations can be very effienct in such computations.. Perhaps more research should go into the analog world before we can begin modeling brains. (The end result would be a hybrid. Something which would not have as reliable a result as a digital computer, but would be faster to make decisions than a human and yet not require such input precision as a computer (i.e. the fuzzy logic of the early 90's)).
The second thing to note is that regular patterns of computational capability are probably not going to be as useful as an irregular array of delay-lined interconnections of CPU nodes. If a sufficient number of CPU's were connected in such a manner, then certain patterns would be exploitable being spacially/temporally repeatable and thus recognizable. Mind you, we'd have to have numbers of CPUs and randomized interconnections on the order of the Human brain. Further, there would have to be a mechanism that allows for interference of data (the key to holography). Perhaps this is the necessary analog stage.
The point isn't really to describe a possible solution, but more to discredit most modern attempts at human consciousness through a computer. Most that I've seen (modeling of a neuron as inductance/capacitance/transistor circuits), don't hit on many of the fundanmental attributes of neural networks, and are thus doomed to failure. Moreover, purely modeling a neuron isn't going to be valuable.. At best, all we'd succeed in is producing a silicon version of a human. But it would necessarily be non-digital in nature. A hybrid is what we really need; taking the best of both worlds, while trying to minimize the worst. For example, hypothetically speaking, we could construct a large irregular array of CPU-nodes (possibly on a single die to be cost effective) that are doing real digital calculations, but the data-sets have interference in a controlled way. Thus when you ask it to divide 242341541234 into 552312341234 it can be as responsive as a modern computer, BUT the act of requesting that information and processing it generates interference for the cognitive portion of the collective node. Thus when instructions are more fuzzy than "please tell me what is x / y" for the above values, the cognitive section can make quick distored decisions of how best to answer the question (e.g. have a truely contextual view of the world). Moreover, the cpu-network dedicated to cognative work would be acting as an overseer of data. Expecting certain patterns. Such activity would be a sort of checks and balances for all processed data.. Theoretically, a simple application would be in determining when/if a cpu node or periferal device has gone bad. The possibilities could be endless, but we're still a while off from the level of technology required to produce billions of randomly interconnected CPU's.
Hopefully... (Score:3, Funny)
Evolution (Score:5, Funny)
Re:Evolution (Score:5, Funny)
1. Build universe's most intelligent computer.
2. Ask for "meaning of life"
3. ???
4. ???
5. ???
9879. Build new computer designed by universe's fastest computer.
9880. Populate with mice.
9881. ???
9882. ???
9883. ???
3228708. Rebuild computer when destroyed by vogons.
3228709. ???
3228710. Profit!!!!
Oh cummon you guys knew it was coming.
No match for me... (Score:5, Funny)
Re:No match for me... (Score:4, Funny)
With that in mind, a P2/266 will probably do it for my brain..
Re:No match for me... (Score:5, Funny)
Looks like it's only a low priority thread.
Brian's nothing special (Score:5, Funny)
Thank god (Score:5, Funny)
Re:Fast, Yes (Score:4, Interesting)
Are we talking about the brain as we use it, or the brain, at it's full potential?
They're the same thing. The brain used the way we use it is the brain.
The idea of a brain that could do a lot more than we ever used it for, by very simple means, is an evolutionary impossibility - it could never have evolved. The idea is absurd.
Processing power of the human brain? (Score:3, Funny)
Actually, that won't be that difficult to do if they are comparing this computer with the "brain power" some of the doe-doe's I went to high school with
Re:Never, EVER more powerful (Score:5, Informative)
That shows that our fingers are slower than our brains. No surprise there.
The fastest computer in the world will always be limited to how quickly data may be fed to it. One way or another, a human will have to direct this operation - if only for safety and security considerations.
That's just silly. Computers can already prcess data much faster than you or I (or you and I) can follow.
not too far away... (Score:4, Interesting)
etc.
How far are we from learning kung fu from an optical disk ?
Whose brain? (Score:3, Funny)
Hooray! (Score:5, Funny)
And it damn better not be 42!
re processing power of the human brain (Score:3, Interesting)
The human brain does more than simple processing. Think about it, the ability to do calculations, etc., is tied into the most ancient (reptilian) part of the brain.
Now, if they could make a computer that could experience emotions (or could explain what women really want :-)), that would be a true accomplishment.
Well - the Orange Catholic Bible says: (Score:5, Funny)
Somehow, I think that might be good advice.
Processing power of the brain? (Score:4, Insightful)
Kargis Strong, MD
Processing Power of the Human Brain? (Score:3, Insightful)
I think that what most people don't get is that the brain is not that powerful a computer... It's just very, very good at what it's supposed to do.
Think of it this way. Instead of a computer and mobo combination, consider the brain as dozens and dozens of embedded micro-controllers that talk to eachother via a protocol. Each one is very specific. We have one that handles getting audio signals, one that handles getting video signals... and then completely different controllers for recognizing voice, music, speech, text, and images. There is one overlying controller-- the frontal lobe-- but most of what is does is pattern matching and random number generation. It's the combination or all these working together, not the raw ability of the brain to process information, that makes the magic of 'conciousness'.
Brain, pfft (Score:3, Insightful)
It's not anything remotely like a human brain. They're making some rough analogy between storage size, processing speed, and the number and nature of neurons in the human skull. This is just a really really really fast/big version of existing machines.
Again, for those who haven't read Douglas Hofstadter's excellent books GEB and MMT - being human-like is a *really* tough thing for a computer, and we haven't even begun to figure out the basics of it on paper. Maybe in 100 years we'll understand the problem better, but I'll place my bets now that when we do we'll finally realize it's futile to try to mimic it.
And they will name it 'skynet' (Score:4, Funny)
Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14am.
Crayola (Score:3, Funny)
I can't wait until a few years from now when we're treated to talking about ASCI Mauve, ASCI Burnt Sienna, and ASCI Periwinkle....
12544 Power5 processors? Damn! (Score:5, Interesting)
I'm sure IBM is working hard on a new interconnect for this beast. Anyone know about the next-generation SP switch?
The press release [ibm.com] also mentions that Purple will consist of "196 seperate computers" -- which works out to 64-processors per computer. Way to go IBM: the current Power4 systems are only to 32-way!
Processing power only part of the issue... (Score:5, Insightful)
The external interfaces are even worse, these make the brain totally useless for many tasks that computers can process in seconds. As an example try raytracing a rendering a scene using crayons and doing the maths in your head.
So the human brain totally and utterly is secondary to the computer already.
Apart from the fact that humans can be inspired. The solution may take a computer 100 years to attack by brute force and it will get there... but a smart person will do it in minutes because "its obvious".
Computers already outstrip us in terms of processing, but while they are just grown up calculators they miss the essence of human processing. A computer hardwired to mutate everything via
It will be generations before computers will have reached a stage they can start doing the obvious. The limited processing of the brain has produced the people on the Jerry Springer show and Isaac Newton, it ain't the hardware, its the software that counts.
Re:Processing power only part of the issue... (Score:5, Funny)
Ah, I beg to differ. Pour orange juice on a motherboard. Totally disfunctional in a few seconds. Now pour orange juice on your head.
Brain 1, Computers 0.
Government brainpower? (Score:5, Funny)
Processing power is not the issue (Score:5, Insightful)
First, the real issue is not hardware or CPU cycles -- it is software. Tired of Seti@home? Let's build a distributed processing network that has as many CPU cycle equivalents as the human brain! Oh yeah, that's already been done. Ok, so why doesn't it "think" yet? Oh yeah...software.
The issue is how to integrate storage, processing, "RAM", etc. into a software package that can emulate a human brain's method of thinking (which may be a very bad, krufty method of developing consciousness -- why would anyone use meat for processors? What a kludgy hack!).
(OT: what if "thinking" software is _not_ GPL'ed? That could be really frightening. So could security issues for "thinking" machines.)
Second, the next issue is why should we compare digital thinking machines to biological ones? Maybe it is the only benchmark we can think of, but given the truly awkward way in which light-sensitive cells were adapted for inclusion a biological thinking machine (see Francis Crick's "Astonishing Hypothesis"), why can't a much more efficient independent decision making machine be developed from digital equipment (not DEC, btw) actually designed for the purpose?
The human brain/computer comparison is really a red herring. The only reason to create a human-like digital thinking machine/emulator (and you thought WINE was hard to use...) might be to pursue immortality. I think the more likely reason is that it would be the ultimate species-wide circle jerk. Humanity getting off on creating humanity. Bleh. Let's set our sights a little higher.
guac-foo
For the 2004 election (Score:3, Funny)
Business Proposal (Score:5, Funny)
I couldn't help but notice that you were hard at work developing a computer to rival the human brain to the tune of $184,000,000.
It just so happens that I have a human brain and I would be quite happy to let you use it for a tidy sum that is far below the aformentioned $184M.
Please give me a call at your earliest convenience to work out the details.
Thanks,
Jason
----[%snip]----
the science of inteligence (Score:4, Interesting)
Although the "rated" processor cycle of a human brain may be measured in Hz... the overall number-crunching and algorithm pattern matching power of 4 billion years of refinement utterly out-class any computer well be making for years to come.
Case in point.. A child playing wall ball makes more physics calculations in one minute of game than a whole team of physicists could map out in months.... he calculates his own mass, his own speed, the angles and exact acceleration of his arms, the weight and distribution of balence between his feet, all while tracking the movements and possible movements of a ball with its own mass and porportions and an opponent. We could count layers upon layers of others things this kid is doing without thought, breathing, processing and responding to components inside his body such as adreneline, and a host of other things... but what it really comes down to is a child's Brain subconsciously is far more powerfull than any comp on the planet.
The comparison of raw number crunching super-clusters to a human who is nearly autonomus, learns independantly and can adapt to many situations in the blink of an eye (where a comp would take considerable reprogramming to adjust to new tasks) is falacy at best.
It has been predicted that AI will reach the emotional awareness of a teenager around 2050
--Enter The Sig
--
There is hope for the scarecrow after all! (Score:5, Funny)
--Joey
He he he (Score:5, Funny)
The Brain: Facts (Score:5, Interesting)
Synapses in adults: 10E14, a few thousand per neuron
Neuron firings per second: max 2 Khz
The biggest challenge in comparing brain to supercomputer is the massive connectivity of brain, with 2000-5000 synapses per neuron.
The total processing speed of ASCII Purple sounds about right for number of neurons in brain times the maximum number of pulses per second per neuron.
Given there are 10E14 synapses, each one with at least a byte of synpatic weight associated with it, it would need memory of at least around a petabyte of memory, although synpase memory change speeds are probably not faster than tape, and I know of plenty of installations with a petabyte on tape.
But here is the kicker: Will those 100 teraflops be flops that can use thousands of inputs? Probably not. So I'd argue that to truly be as powerful as the human brain, you would need 100 petaflops of 1-2 input flops, with at least a petabyte tape system.
Power 5 for my Nintendo?!? (Score:3, Insightful)
Umm, how about...NOT. Just because they're all PowerPC based doesn't make them the same. Based on that logic a 386 and a Pentium 4 are the same too, just beacuse they're both built on the x86 architecture.
Power 5 (can't find a link) is a generation of chips that are related, but further on the horizon than the chips Apple is buying [slashdot.org] (both are Power 4 spin-offs, but quite different). The chips used in the Nintendo GameCube [ibm.com] are not even related -- they just happen to also be made by IBM -- not to mention they are several years old while the above chips are not even available yet.
Then again having a server class chip in a Nintendo might be interesting...
BS (Score:3, Insightful)
btw, I'm a ee who does neuroscience research, so I'm not talking out of my ass here.
currency is off. (Score:3, Informative)
Re:anyone else tired of "boom boom" computers? (Score:4, Insightful)
>
> Considering we can blow up the surface of the world a couple of times(at least) over with our existing stockpiles, why are we spending ANY money on ANYTHING except REDUCING said stockpiles?
It's you :-)
Seriously - reducing the need for large nuclear stockpiles exactly why the money's being spent on simulations.
Nukes are complicated devices, composed of weird stuff (the fissionables and other what-not), and normal stuff (the explosives that trigger the weird stuff).
Over time, the weird stuff changes its properties. So does the normal stuff.
One of many issues with nukes is that if you're gonna throw one at someone, you want to be damn sure it goes off. Otherwise, you've probably just given your enemy enough weird stuff that they could build their own bomb. This, I think we can agree, is a Bad Thing.
If you're going after a guy in a hardened bunker, and your nuke blows up but doesn't blow as strongly you thought it would, you may have to lob another one at the same target. And that means you need to have more nukes in reserve.
And worse yet, if you're going after the same bunker, but your nuke works a little too well, you've just wiped out a city instead of just the few hundred feet around your target. This is inefficient at best, and barbarism at worst. (The early fusion bombs had this "problem", and some tests resulted in radiation exposures far greater than was expected, mainly because the bomb was "better" than it was supposed to be.)
If you want to cut down on the number of nukes in the arsenal, a good way is to make sure that you've got a few very good ones that always go off when they're supposed to, with the correct amount of "boom".
One way to make damn sure your nukes blow up when and how big they're supposed to is to test them regularly. I'll grant that mushroom clouds over the Nevada desert were probably very pretty to watch, but they were also pretty messy for those living downwind. Bad idea.
The second way is underground testing, which solves most of the "downwind" problem, but can still result in some leakage under some circumstances.
That really only leaves one other option - to run simulations. Lots of simulations. Using the best math your scientists can come up with, and the fastest computers your geeks can build. No radiation leaks, and what you learn while building the supercomputers can be used for building higher-performance computers for peaceful purposes in the future.
I dunno about you, but I'll take Door Number Three any day.
What OS? (Score:3, Interesting)
stupid name for a supercomputer.... (Score:3, Funny)
ASCI Purple and Blue Gene/L ? WTF?
Is the next version going to be called
ASCI Pink and Purple? or ASCI Barbie's Dreamhouse.....
Get back to naming the systems after Tolkien characters, or greek gods. (
Skynet will rule the human race, sure enough, but it won't be called "Skynet". It will be known as
ASCI SuperPoopyPants.
nbfn
Mike Nelson? (Score:3, Funny)
Shouldn't Joel Robinson be the director of this project? I mean, the guy made at least three AIs out of parts meant to stop and start movies! Mike was barely able to keep them functioning after Joel escaped.