• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Simulating one second of real brain activity : 40 minutes, 83K processors, 1PB of RAM

Status
Not open for further replies.

livestOne

Member
A team of Japanese and German researchers have carried out the largest-ever simulation of neural activity in the human brain, and the numbers are both amazing and humbling.

The hardware necessary to simulate the activity of 1.73 billion nerve cells connected by 10.4 trillion synapses (just 1 percent of a brain’s total neural network) for 1 biological second: 82,944 processors on the K supercomputer and 1 petabyte of memory (24 bytes per synapse). That 1 second of biological time took 40 minutes, on one of the world’s most-powerful systems, to compute.

If computing time scales linearly with the size of the network (a big if; I have no idea if this would be the case), it would take nearly two and half days to simulate 1 second of activity for an entire brain.

the first A.I. with only 1 second to think

w͎h͓̝̞̮͍͎̋̿ͫë̟̞́ͯ̋̊rͣ̈̂̂̎͛e̲̝̱͈͋̄̒̉̇ ̩͓ͩ̉̿̾ͬ͊a̘͔̲ͨͥͩ̊m̗̝̺̺̜̙̻ͯ̐ͨͫ̊͐ ͎͈͉̳̤͓͎͂̀ͣ̊̚I̮͍̙͎̗͓ͫ͆͛ͅ?̰̗
 

DopeyFish

Not bitter, just unsweetened
We are very, very far away from creating a conscious machine. I wouldn't start worrying about that just yet.

Well... At least through simulation/emulation (piece by piece)

The trick is entirely in the consciousness, something no one knows how to construct (at the moment)

Another problem is... How much time would we need to be sure we even made one? How much pre-built functionality do we have as humans that would need to be compensated for in an artificial being?

And that's even before you figure out how to inject sensory systems within the consciousness itself.
 

Log4Girlz

Member
Scientists simulated in 10 seconds, 3 biological weeks of my brain activity with a gameboy color and 1 mb of storage space.
 
We are very, very far away from creating a conscious machine. I wouldn't start worrying about that just yet.
Indeed we are, that's part of what I was trying to get across. But you have to start worrying at some point, right?

You're assuming quite a lot here. Most importantly, those views assume a purely naturalist premise as well as side-stepping the hard problem of consciousness.
Assuming a lot, yes, but not necessarily side-stepping. The hard problem actually works in favour of these views. Since we can't grasp the very essence of conciousness, our best bet is to either assume 1) it doesn't really exist or 2) it lies in the activity itself. Both of these imply that a simulation would effectively have it on the same level as we do.

(I know I'm full of shit, I have different names for the concepts I had to look up from your post)
 
w͎h͓̝̞̮͍͎̋̿ͫë̟̞́ͯ̋̊rͣ̈̂̂̎͛e̲̝̱͈͋̄̒̉̇ ̩͓ͩ̉̿̾ͬ͊a̘͔̲ͨͥͩ̊m̗̝̺̺̜̙̻ͯ̐ͨͫ̊͐ ͎͈͉̳̤͓͎͂̀ͣ̊̚I̮͍̙͎̗͓ͫ͆͛ͅ?̰̗

I must know how
 
M

Macapala

Unconfirmed Member
7jYzls2.jpg
 

sirap

Member
They could have achieved this faster if they hooked up to the cloud.

You're crazy, that's like flipping the switch on Skynet. It's bad enough that we have it watching us silently through Kinect, now you want to give it intelligence?!!
 
sounds like bullshit. i'm sure that we can't even simulate one eukaryotic cell because we don't even know how they work, let alone have enough processing power to model that many atoms.
 
think about how good your brain is at rendering your dreams, bro
I don't know, my brain has really poor fill rate. Actually there doesn't seem to be any rasterization; I'm not getting any pixels in frames, more like some weird conceptual sensations. NPC AI has no consistency and the whole thing crashes quite often. And I don't even know how much real-world time I actually spend processing the perceived dream time.

It's hard to form a valid analogy. No matter how vivid worlds we experience in dreams, we certainly can't output them 1920x1080@60. The whole pipeline from imagination to presentation should be looked at for comparison with computers.
 
Go science!

Science didn't do anything. This experiment was done by researchers from the RIKEN HPCI Program for Computational Life Sciences, the Okinawa Institute of Technology Graduate University (OIST) in Japan and Forschungszentrum Jülich in Germany. They didn't even get a proper credit in that article.
 

davepoobond

you can't put a price on sparks
And yet humans are still such dumb fucks.

.


the obvious solution is not to build a mechanical object that can do the same calculations as a brain, it is to harvest brains and use them AS COMPUTERS


Science didn't do anything. This experiment was done by researchers from the RIKEN HPCI Program for Computational Life Sciences, the Okinawa Institute of Technology Graduate University (OIST) in Japan and Forschungszentrum Jülich in Germany. They didn't even get a proper credit in that article.



BUT.....buttttt....SCIENCE!!!!!!!!!!!!!!!!
 

The Technomancer

card-carrying scientician
Massively parallel networks like this are one of a couple of practical areas of computing that quantum computing could seriously improve. But we're decades away from building a machine capable of doing such a simulation.

Sure, but what is the architecture of this current computer they're talking about? I'm saying that I don't think the register-RAM-instruction model is ideally suited to modeling the brain, and I'm wondering what kind of fundamental re-designs we could do still just using transistors.
 

Ether_Snake

安安安安安安安安安安安安安安安
Singularity is going to happen a week after he dies.

He's going to have his brain frozen or some such.

But will then be brought back to life 100 years from now for a reality to show where rich people who froze their brains in the 21st century are brought back to life by having their brains implanted in machines that fight one another as gladiators, to everyone's amusement.
 

Zaptruder

Banned
Brain architecture is more reminiscent of a massive network of computers than it is of an actual computer.

That is to say, each neuron is its own processing machine that processes the incoming signals and routes the connection data to the next machine that relates to the processed pattern of information.

At the level of the neuron... it's basically a pattern information processor - a series of electrical on and offs that it relays to a connected partner, which in turn finds out which partner is best to send that information to (or multiple partners).

The unique temporal pattern of traversal and activation of mass networks of neurons is what formulates the sensation of conciousness that the overall neural network experiences. This is similar to saying the unique temporal pattern of on and offs in a CPU creates the end product of what we see and interact with on the screen.

Another way to describe the makeup of the brain is to say that it is a dynamic CPU that is able to rearrange its configuration... the hardware and software is intertwined - and thus able to do much more complex computation with much less.

This kind of research won't be used to replicate human-like brains in the long run... instead there'll be much more useful for us in understanding the subtleties of what makes brains such great information processors - and how those strengths can be redesigned in a manner that can usefully be replicated in silicone or other non-organic elements.
 

RyanDG

Member
But will then be brought back to life 100 years from now for a reality to show where rich people who froze their brains in the 21st century are brought back to life by having their brains implanted in machines that fight one another as gladiators, to everyone's amusement.

I would pay good money to watch this movie.
 

Diablos

Member
The brain ain't got shit on my i5! I'll just slap some Arctic Silver 5 on there, OC to 4.4GHz and I'll have a brain in my box in no time! Thanks newegg!
 
Why do I struggle in Math then?

Because your brain sucks at it. Most of your brain is specialist hardware for doing one thing or another. It's fantastic at language and facial recognition but abysmal at mathematical operations. High level reasoning, planning, maths etc are not as computationally intensive as you might think, while getting something to stand upright or parse a sentence is much harder than you might think. Yet we are designed to do the latter, so it seems trivial to us, while we are not designed to do the former so it seems really hard or complicated. This is known as "Moravec's paradox".
 

wsippel

Banned
K-supercomputer may not be the best choice for the task. 82,944 cores sound impressive, but it can fit on 31Tesla cards , just saying.
You'd need more than 1,000 Tesla cards to achieve the same performance. K is the fourth most powerful supercomputer. The Cray Titan with ~300,000 Opteron cores and ~19,000 Tesla cards is only about twice as fast. K actually has 705,024 cores, this experiment didn't use the whole thing.
 

sirap

Member
Because your brain sucks at it. Most of your brain is specialist hardware for doing one thing or another. It's fantastic at language and facial recognition but abysmal at mathematical operations. High level reasoning, planning, maths etc are not as computationally intensive as you might think, while getting something to stand upright or parse a sentence is much harder than you might think. Yet we are designed to do the latter, so it seems trivial to us, while we are not designed to do the former so it seems really hard or complicated. This is known as "Moravec's paradox".

Interesting. The geniuses that are born with wizard math skills lack the opposite.
 

Wellscha

Member
Because your brain sucks at it. Most of your brain is specialist hardware for doing one thing or another. It's fantastic at language and facial recognition but abysmal at mathematical operations. High level reasoning, planning, maths etc are not as computationally intensive as you might think, while getting something to stand upright or parse a sentence is much harder than you might think. Yet we are designed to do the latter, so it seems trivial to us, while we are not designed to do the former so it seems really hard or complicated. This is known as "Moravec's paradox".

Interesting stuff.
 
Because your brain sucks at it. Most of your brain is specialist hardware for doing one thing or another. It's fantastic at language and facial recognition but abysmal at mathematical operations. High level reasoning, planning, maths etc are not as computationally intensive as you might think, while getting something to stand upright or parse a sentence is much harder than you might think. Yet we are designed to do the latter, so it seems trivial to us, while we are not designed to do the former so it seems really hard or complicated. This is known as "Moravec's paradox".

I feel like I have stepped inside a wonderful edited wiki page.
 
He's not a scientist,

from wikipedia

Kurzweil was the principal inventor of the first CCD flatbed scanner,[2] the first omni-font optical character recognition,[2] the first print-to-speech reading machine for the blind,[3] the first commercial text-to-speech synthesizer,[4] the first music synthesizer capable of recreating the grand piano and other orchestral instruments,[citation needed] and the first commercially marketed large-vocabulary speech recognition.[5]
 
Status
Not open for further replies.
Top Bottom