IBM Takes a (Feline) Step Toward Thinking Machines

bth writes "A computer with the power of a human brain is not yet near. But this week researchers from IBM Corp. are reporting that they've simulated a cat's cerebral cortex, the thinking part of the brain, using a massive supercomputer. The computer has 147,456 processors (most modern PCs have just one or two processors) and 144 terabytes of main memory — 100,000 times as much as your computer has."

Comments

news for nerds

[blitzkrieg3]

(most modern PCs have just one or two processors)

Aren't we expected to know that? This is /. after all...

Re:news for nerds

[visualight]

Yes it does, don't be pedantic.

Re:news for nerds

[Hal_Porter]

Yes, yes it is.

http://en.wikipedia.org/wiki/Symmetric_multiprocessing

In computing, symmetric multiprocessing or SMP involves a multiprocessor computer architecture where two or more identical processors can connect to a single shared main memory. Most common multiprocessor systems today use an SMP architecture. In the case of multi-core processors, the SMP architecture applies to the cores, treating them as separate processors.

You disagree with Wikipedia. That means you've been proven wrong in front of the whole Internet. Hang your head in shame.

Re:news for nerds

[nschubach]

I was trying to figure out who they were talking about when they said "your computer." ;)

The review looks like it was written for a grade school presentation with that and the processor comment.

Re:news for nerds

[Chris Burke]

Technically is a single Core2Duo/Quad or Core iX CPU considered SMP? I would guess no they are not.

Funnily enough, a single Core i7 or Opteron is SMP, but if you have multiple, then it isn't, it's NUMA because not all the processors have Symetric access to memory.

Core 2 is SMP for all standard configurations.

Re:news for nerds

[MaskedSlacker]

What I want to know is how is the fact that you guessed supposed to make your erroneous spouting better?

When you spout misinformation, which is a serious problem on the internet (i'm looking at you conspiracy theorists), the fact that you guessed doesn't absolve you, since 15 seconds effort on your part would have meant 1 less piece of misinformation forever preserved.

I propose that you sir are an internet asshole (not that this particular piece of misinformation means a goddamn thing, since everyone who reads /. knew you were wrong, but THINK OF THE KITTENS!)

Point is, misinformation is a problem here, and being glib about the fact that you're a lazy trollop makes it worse, not better.

Cool...

[Blazarov]

Does it keep wanting cheezburgerz all the time?

Re:Cool...

[bobdotorg]

I can has petaflop?

Re:Cool...

[ImABanker]

Does it dream of electric mice?

Re:Cool...

[Blakey Rat]

I doubt IBM is up on Internet memes enough to get that one. I'd wager they're still laughing at the Ally McBeal dancing baby animation.

Well I hope

[Cornwallis]

the first thing they teach it is to stop scratching my couch.

One word...

[wfstanle]

One word...

Meow!

hmmmm

[Polkyb]

They've spent millions teaching a computer how to destroy furnature and shit in your shoes.

Re:hmmmm

[LWATCDR]

Funny I was going to say lay in the sun and ignore you.

Re:hmmmm

[ArcherB]

Funny I was going to say lay in the sun and ignore you.

It's IBM,not Sun.

Sleep Mode

[Anonymous Coward]

Now if the could just get it out of sleep mode.

Re:Sleep Mode

[Dannon]

Just move the mouse.

Re:Why cats?

[Anonymous Coward]

To properly simulate a cat, the system needs to sleep 20hrs a day which would thus make it the greenest option on the market.....

But....

[jimicus]

Can it lick its own arse in polite company?

meat versus silicon and metal

[jollyreaper]

It amazes me how much hardware and power has to be thrown at the problem to solve it while nature can create a self-organizing machine that only requires material input of raw mice and lasagna. Puts me in mind of this quote:

"If research leads to the development of successful new modeling techniques that can carry out new and better forms of information processing, no one will really care if they do not exactly mimic the functionality of the human brain," concludes Hall. "I honestly doubt you'll find too many people today who are upset that the wings on an aircraft do not flap like those of a bird or that a submarine does not swim exactly like a fish."

It's an interesting way of looking at things. Man's earliest ideas of flying all involved trying to mimic the actions of a bird. And ornithopters remain impractical as passenger vehicles. But new breakthroughs in material sciences and computing are allowing for autonomous bots that fly like birds, bats, bugs, and can swim like snakes and fish. Engineers will point out that the evolved solutions we see in nature are working with the materials at hand, they might not be the best of all solutions. Every flying vertebrate known to science turned forelimbs into wings and flap them. Is it the most efficient way to fly? That's an argument I'll leave to the biologists and engineers but it's certainly the only way those vertebrates were getting into the air! They have to work with the materials at hand. If we ever saw flying horses, the only thing we could be absolutely sure of is that this would not be achieved by sprouting two more limbs from the back. We see evolution taking away limbs but never adding new ones.

Re:meat versus silicon and metal

[quadrox]

Uh yeah, because evolution started with creatures that had 4 limbs and 5 toes/fingers on each, right? These didn't evolve over time, right?

I'm sorry, but you are wrong for obvious reasons.

Re:meat versus silicon and metal

[agrif]

I promised myself I wouldn't be a quote-quoter, but really, you guys make it too easy. The quote above from Hall most likely references this, from one Edsger Dijkstra:

“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.”

Unfortunately, you'll find a lot of people that think he meant "Submarines don't swim, you retard! So computers don't think!" It seems pretty clear to me that he means making computers think like organisms would be an inefficient and pointless gesture, as they are capable of something far less primitive.

(I found this quote in Accelerando, by Charles Stross, and loved it. It's Creative Commons, so you have no excuse not to read a little.)

Iz in ur brane...

[benwiggy]

Iz in ur brane, making ur thorts. LOL!

"The computer has 147,456 processors and 144 terabytes of main memory."

First there was "Deep Thought"

[Average_Joe_Sixpack]

then "Deep Thought II"
then "Deep Blue"
next "Deep Pussy"??

Nah, but it will refuse to be mouse operated

[Anonymous Coward]

Nah but it will refuse to be mouse operated ...

The Paper

[glwtta]

Here's the actual paper (pdf).

Although, of course, posting the piece of pap that explains how many processors my machine has makes so much more sense.

Wasn't Slashdot supposed to be for a semi-technical audience? Hell, even a semi-literate one.

They are a model organism for neuroscience

[tpjunkie]

Having done neuroscience research, (if only on a master's degree level), I can say that the cat brain is particularly well studied, mapped out, and understood by neuroscientists. It is used as a model organism by many neuroscientists, and has a number of similarities with the human brain in its layout and function, much moreso than the mouse or rat brain.

Re:They are a model organism for neuroscience

[Kell Bengal]

Surely there is an easier way for computer nerds to get pussy?

A pile of neurons does not a brain make...

[swm]

From TFA, it doesn't sound like they simulated the cerebral cortex of a cat.
It sounds like they simulated a neural net with a comparable number of neurons.
Not the same thing.

A few days ago, Slashdot ran The Math of a Fly's Eye May Prove Useful.

Those guys

• reverse engineered the yaw motion detector in a fly brain
• reduced the neural network to a set of 5 coupled, non-linear equations
• implemented the equations on a computer
• ran their implementation against an animated scene
• observed that the equations correctly and robustly detect yaw

and they still don't understand how the equations actually work.

That's where we are with brain simulation.

Re:A pile of neurons does not a brain make...

[Pedrito]

From TFA, it doesn't sound like they simulated the cerebral cortex of a cat. It sounds like they simulated a neural net with a comparable number of neurons. Not the same thing.

What article did you read? The one linked to in the post clearly says they simulated a portion of cat cortex and, in fact, that's largely what they did. There's more here about some of the specifics. It's not an entirely accurate simulation, but it's pretty close. Not all neuron types are represented and it's largely cortical, thalamus and reticular nucleus neurons. They've created cortical hypercolumns which is the way a real cortex is laid out. They've omitted the layer 1 neurons, but otherwise the cortex is probably pretty functional for what they're doing. I think it's a pretty amazing feat.

Re:A pile of neurons does not a brain make...

[jcaplan]

TFA is bunk. (Yes, I read it.) 12 pages of bunk. Much of the article is about the computational challenges and blathers on about number of processors used and memory. Under key scientific results, they find that their model propagates waves at about the same rate as is found physiologically. So they connected a bunch of nodes in a way that produced synchronous behavior at a certain frequency. I could tune any model you give me to produce this behavior. (I have no special talent here, anyone writing models could.) Yawn. They ramble on about signals propagating between layers at reasonable rates, too. And ...?

What about their simulation doing anything like what a cat might naturally do, such as detect a moving object? Nope. Instead they go on to discuss the scaling of their model, profiling and performance modeling. Perhaps one reason their model shows absolutely nothing is that they have connected their simulated neurons randomly. Yes. Randomly. Or as they put it: "The coordinates of target thalamocortical modules for each cell are determined using a Gaussian spatial density profile centered on the topographic location of the source thalamocortical module". Yep, thats random. Since their model doesn't ever change connection strengths (one form of learning) these random connections never change.

I recently heard a description of the ways you can fool someone with computational neuroscience. Here are a couple of them: "Two card monte" Write a paper that spans two fields, but has no significant results in either. The specialists in one field will feel that the work done in their field is trivial, but that exciting stuff from the other field in the paper is what makes it so special. The specialists from the other field may feel the same way. Somebody snookered the conference organizers into thinking they were doing any neuroscience at all. The other was called "Turning the prayer wheel" or burning compute cycles to gain scientific merit. Fancy hardware is cool, but it can produce absolutely trivial results as this paper confirms.

I don't mean to say that this research is entirely pointless. Indeed it has succeeded in siphoning significant funding from DARPA which might otherwise have gone into developing [killer] robot dogs.

Cat-Brain Tech

[LaminatorX]

The military is rumored to be interested in using the cat simulator to guide precision munitions with laser pointers. Unfortunately the system seems limited to short range applications, as missiles seem to loose interest after a couple minutes.

cat-SIZED brain, not a cat brain

[WAG24601G]

This project is basically a massive neural network simulation with a number of nodes and connections comparable to the estimated totals in a cat's brain. In short, there is nothing cat-like about this system apart from its raw processing power.

Not to reduce the value of this feat, by any means! There are tons and tons of neural network simulations that can produce roughly human-like results in very, very narrow domains, but as the quote below explains, these simulations are decades (or more) from connecting the behavior of tiny subsystems (a few hundred neurons) with the overall phenomenon of 'mind' (conscious and unconscious cognition). The expectation is that a network of this size will show some new emergent properties that will give us clues about the intermediate "higher than cells, lower than interviewing a human" order of processing.

Jim Olds, a neuroscientist and director of the Krasnow Institute for Advanced Study at George Mason University, called the new research a "tremendous step." Olds, who was not involved in IBM's work, said neuroscientists have been amassing data about how the brain works much like "stamp collectors," without a way to tie it together.

"We've made tremendous advances in collecting data, but we don't have a collective theory yet for how this complex organ called the brain produces things like Shakespeare's sonnets and Mozart's symphonies," he said. "The holy grail for neuroscientists is to map activity from single nerve cells, which they know about, into how billions of nerve cells act in concert."

Re:"100,000 times as much as your computer has"

[Anonymous Coward]

It's often thought that gibibytes and tebibytes were invented to allow "giga" and "tera" to retain their conventional meanings as powers of 10 even when used to refer to quantities of data.

However, the true reason was to enable an entirely new form of pedantry.

Re:Cat mentaity

[Dannon]

Sounds like Windows ME.

Windows MEow in this case.

Aineko? Is that you?

[molo]

This reminds me of Aineko in Accelerando by Stross. I wonder how long until it becomes sentient and surpasses human intelligence. See here: http://en.wikipedia.org/wiki/Accelerando_(novel) http://www.accelerando.org/

-molo

That's easy

[Minwee]

I did something similar this morning.

while true ; do
echo "I hate you."
echo "Feed me."
sleep 60
done

Now how hard was that?

Re:"100,000 times as much as your computer has"

[cmiller173]

gibi/giga whatever, I'm pretty sure that this cat simulating computer actually has a bazillion kibble-bits.

Simulation output

[nodrogluap]

My captors continue to torment me with bizarre dangling objects. They eat lavish meals in my presence while I am forced to subsist on dry cereal. The only thing that keeps me going is the hope of eventual escape... that, and the satisfaction I get from occasionally ruining some piece of furniture. I fear I may be going insane.

How hard is it to simulate a cat's brain on 1 cpu?

[tomhudson]

simulated a cat's cerebral cortex, the thinking part of the brain

10: INPUT(8) $SOUND
30: IF ($SOUND == 'CAN OPENER') GOTO 140
40: DO CASE (RND(4))
50 CASE 1:
60 CLAW_FURNITURE()
70 BREAK
80 CASE 2:
90 MARK_FURNITURE()
100 BREAK
110 CASE 3:
120 SLEEP(RND(10000))
130 CASE 4:
140 PRETEND_TO_BE_NICE()
150 IF (FOOD) EAT()
160 GOTO 10
170 ENDCASE

I already have it on my computer!

[maxwell demon]

I just looked into my /bin directory, and there it was: An executable clearly named "cat"!

Has some biological properties

[mmacdona86]

Reading the TFA, it looks like they went to some trouble to model some specific brain structures and synapse properties, including inter-area connectivity and learning, in the model. So it's not "Just a big neural net." However the accuracy of the simulation is limited--both by what we know about the detailed structure of the cat's brain and by the number and complexity of the structures they decided to model.

Cores does not equal processors

[gyrogeerloose]

For what it's worth, here's a text dump from the Apple System Profiler on my MacBook Pro:

Model Name: MacBook Pro
Model Identifier: MacBookPro1,1
Processor Name: Intel Core Duo
Processor Speed: 2 GHz
Number Of Processors: 1
Total Number Of Cores: 2

So, it would appear that Apple at least does not equate the number of cores and the number of processors.

Iz in yur timez...

[DarthVain]

...termanatin yur connerz.