The biggest and most promising project to reverse-engineer the brain on an IBM supercomputer

[johnnd]

Henry Markram: Simulating the Brain — The Next Decisive Years [1/3]

 

Henry Markram: Simulating the Brain — The Next Decisive Years [2/3]

 

Henry Markram: Simulating the Brain — The Next Decisive Years [3/3]

 

 

 

 

Henry Markram, Ph.D., Director of the Blue Brain Project at École Polytechnique Fédérale de Lausanne, speaks at the International Supercomputing Conference 2011.

 

 

Not sure where to post this, but this is a giant engineering project, is it not?

 

 

[Enthalpy]

What a hype!

 

Mimicking the brain on a computer has been attempted so many times, including on the biggest supercomputers, and on specialized machines...

 

Is there any paper around here that tells what could make this project different from the previous and ongoing ones?

 

From what I read they use banal processors as everyone does. In big quantity, mutitasked, and with a connection network - again as every big computer. They tried to contain the power consumption through the choice of the processors, as has been done before as well, and doesn't change so much.

 

To the very least, what is the connexion network they use? I mean, the electronics network between the processors - not the synapses they want to emulate.

[swansont]

Not sure where to post this, but this is a giant engineering project, is it not?

 

!

Moderator Note

Spamming it all over the forum was not the right answer. Duplicates have been deleted. One thread to a topic, please

[johnnd]

Mimicking the brain on a computer has been attempted so many times, including on the biggest supercomputers, and on specialized machines...

 

It really hasn't. If you're talking about simple point neuron simulations, then fine. But the goal of the Blue Brain Project is to faitfully follow real biology, not make theories and models. Besides, we haven't even had the requisite computing power up to this point to simulate the brain. Even assuming relatively simple neural complexity, we can estimate the computing power of the brain to equal 10 ^ 12 neurons * 1000 synapses/neuron * 100 Hz operation/neuron = 10 ^ 16 operations per second. The most powerful supercomputer in the world has about this power. However, Markram estimates that a one exaflop machine is needed for full-brain simulation, which will not be available until 2018 at the earliest. That's why the deadline for the project is in 2023.

 

 

 

[dragonstar57]

It really hasn't. If you're talking about simple point neuron simulations, then fine. But the goal of the Blue Brain Project is to faitfully follow real biology, not make theories and models. Besides, we haven't even had the requisite computing power up to this point to simulate the brain. Even assuming relatively simple neural complexity, we can estimate the computing power of the brain to equal 10 ^ 12 neurons * 1000 synapses/neuron * 100 Hz operation/neuron = 10 ^ 16 operations per second. The most powerful supercomputer in the world has about this power. However, Markram estimates that a one exaflop machine is needed for full-brain simulation, which will not be available until 2018 at the earliest. That's why the deadline for the project is in 2023.

how can people say that the brain is faster than a computer. my computer seems smarter than many people i know. it never spells things wrong and does math problems almost instantly etc

[johnnd]

how can people say that the brain is faster than a computer. my computer seems smarter than many people i know. it never spells things wrong and does math problems almost instantly etc

I don't think there is any question that computers operate at tens of millions times the speed of biological neurons. Also, they can instantly remember billions of facts, something that humans would be hard-pressed to do. Where humans excel is at massively parallel processing, which is good for things like language or recognizing faces, or a song, or a complex idea.

 

The argument which I don't buy is that because computers are below human levels today (by virtue of not having designed to work like humans up to this point), they will never reach or soar past that level. The jury is still out.

 

[Enthalpy]

Self-quoting Enthalpy: "Mimicking the brain on a computer has been attempted so many times"

 

It really hasn't. If you're talking about simple point neuron simulations, then fine.

 

Yes it has. I helped a colleague in 1985 who was doing such things and they weren't new at that time. Certainly not a single neuron.

 

Newspapers don't take the time to study a topic before reporting about it, but you can do it.

[johnnd]

Self-quoting Enthalpy: "Mimicking the brain on a computer has been attempted so many times"

 

 

 

Yes it has. I helped a colleague in 1985 who was doing such things and they weren't new at that time. Certainly not a single neuron.

 

Newspapers don't take the time to study a topic before reporting about it, but you can do it.

I did not say 'single neuron', I said 'simple neuron'. Certainly in 1985, the complexity of such neurons was nowhere near the level of complexity in Markram's simulations. Back then, we couldn't really peer into neurons the way we do today. I'm sure your colleague wasn't mathematically modelling the morphologies of hundreds of types of neurons, ion channels, etc.

 

Complexity DOES matter. Consider this: It took Markram's team an IBM supercomputer to simulate a single cortical column with 10,000 neurons. In 1985, a similar machine would have had on the order of tens of millions of times less computing power, not enough to simulate a single neuron.

[Virtual Brain 2045]

Johnnd, I so agree with you, I follow this amazing project for several years now and it's just amazed me.

 

I can't see any reasonable reason why we will not be able to build thinking machines that have consciousness just like we have.

 

Our Brain is the Ultimate Proof that it is possible, there are no magics or mystics in the brain, and if a totally blind brainless process like the evolution could create it, then I have no doubt in my mind that with our advanced technology, and our intelligent and our creativity we will be able to create something at least at this level, and probably much more.

 

We are living in an amazing time, no doubt about that, more and more things and abilities that considered to be science fiction just a few years ago are now becoming reality just in front of our eyes:

 

Digital cameras that have the ability to recognize people faces, computers that can detect objects in a picture (or film) and say, here there is a cat, here we see a house, this is a tree... they can also tell now the difference between a cat and a dog!

 

http://pascallin.ecs...2010/index.html

 

http://cg.cs.tsinghu...ontage/main.htm

 

 

something that lot of people claimed that computers will Never be able to do. Computers can automatically turn human speech into text (available in most YouTube videos) they can also drive cars (search for Google self driving car) , flying airplanes (search for "X-47B") and they can also Defeat Humans on an intelligent competition -

 

 

 

See also this 2 great lectures of Henry Markram:

 

http://ditwww.epfl.c...=10&win_close=0

 

http://www.neuroinfo...arkram.flv/view

 

 

I say that it's possible, and will be done in the next 25 years.

 

....

 

Check also this:

 

 

....

Ray Kurzweil on Glenn Beck

 

1. http://www.youtube.c...h?v=dXvo68Sf-no

 

2. http://www.youtube.c...h?v=4JT2aDPVOEE

 

3. http://www.youtube.c...h?v=gRT7VXRTaI4

 

4. http://www.youtube.c...h?v=ej91NYwpXhg

 

 

....

IBM builds cognitive chips that emulate the human brain

 

http://www.vision-sy...uman-brain.html

 

http://www.engadget....human-brain-her

 

....

Holograms Reveal Brain's Inner Workings

http://actu.epfl.ch/...-inner-workings

 

....

Extract images directly from brain:

http://pinktentacle....ctly-from-brain

 

....

Monkey Moves Robot Using Only His Mind:

 

....

Fully Integrated Modular Prosthetic Limb:

 

....

The Cyborg professor:

http://www.cnn.com/2...borg/index.html

 

....

I am my connectome:

 

 

.

Edited September 10, 2011 by Virtual Brain 2045

[Virtual Brain 2045]

 

I added some links, and fixed some that didn't work.

 

Check them out, they are real good.

 

 

 

 

[johnnd]

Thanks Virtual Brain, I've seen most of those links already. Here are a couple more:

 

Blue Brain: Year One

Blue Brain: Year Two

 

 

It's a ten-year film in the making.

Edited September 19, 2011 by johnnd

[matty]

Speaking of supercomputers, I wonder if the subject of this: http://news.bbc.co.uk/2/hi/technology/5287254.stm has ever come up in here, stumbled on it, though it's old, while trying to reference a recent article from mos ago on protein folding, sparked by a thread floating around in here somewhere.--But kinda cool, don't remember if that thread mentioned gaming-for-a-cure, since I can't relocate it as yet...

Edited October 10, 2011 by matty