Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> Off Topic Discussion >> wow, a neuronal net outgrowing a brain's cat
        
(Message started by: clauchau on Nov 19^th, 2009, 8:58am)

Title: wow, a neuronal net outgrowing a brain's cat
Post by clauchau on Nov 19^th, 2009, 8:58am

IBM and Stanford University just announced they have built and run a simulated brain about the size and connectivity of a cat's brain, except it runs 500 times slower. They expect to reach the size and real speed of the human brain in 2019!

http://p9.hostingprod.com/@modha.org/blog/2009/11/post_3.html

Quote:

The model reproduces a number of physiological and anatomical features of the mammalian brain. The key functional elements of the brain, neurons, and the connections between them, called synapses, are simulated using biologically derived models. The neuron models include such key functional features as input integration, spike generation and firing rate adaptation, while the simulated synapses reproduce time and voltage dependent dynamics of four major synaptic channel types found in cortex. Furthermore, the synapses are plastic, meaning that the strength of connections between neurons can change according to certain rules, which many neuroscientists believe is crucial to learning and memory formation.

At an anatomical level, the model includes sections of cortex, a dense body of connected neurons where much of the brain's high level processing occurs, as well as the thalamus, an important relay center that mediates communication to and from cortex. Much of the connectivity within the model follows a statistical map derived from the most detailed study to date of the circuitry within the cat cerebral cortex.

We are able to observe activity in our model at many scales, ranging from global electrical activity levels, to activity levels in specific populations, to topographic activity dynamics to individual neuronal membrane potentials. In these measurements, we have observed the model reproduce activity in cortex measured by neuroscientists using corresponding techniques: electroencephalography, local field potential recordings, optical imaging with voltage sensitive dyes, and intracellular recordings. Specifically, we were able to deliver a stimulus to the model then watch as it propagated within and between different populations of neurons. We found that this propagation showed a spatiotemporal pattern remarkably similar to what has been observed in experiments with real brains. In other simulations, we also observed oscillations between active and quiet periods, as is often observed in the brain during sleep or quiet waking. In all our simulations, we are able to simultaneously record from billions of individual model components, compared to cutting-edge neuroscience techniques that might allow simultaneous recording of a few hundred brain regions, thus providing us with an unprecedented picture of circuit dynamics.

I love it. However I'm concerned they suggest they can do whatever they want with it without ethical questions, unlike with a real cat's brain.

Title: Re: wow, a neuronal net outgrowing a brain's cat
Post by omar on Nov 24^th, 2009, 5:12pm

Interesting news. Thanks for sharing that Claude.

Title: Re: wow, a neuronal net outgrowing a brain's cat
Post by Janzert on Nov 24^th, 2009, 8:30pm

And the rebuttal? angry rival?

Cat Fight Brews Over Cat Brain (http://spectrum.ieee.org/blog/semiconductors/devices/tech-talk/blue-brain-project-leader-angry-about-cat-brain?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+IeeeSpectrum+%28IEEE+Spectrum%29)

Janzert

Title: Re: wow, a neuronal net outgrowing a brain's cat
Post by Fritzlein on Nov 25^th, 2009, 6:59am

Thanks for the followup link, Janzert. It had been my understanding that simulating even one neuron was a tricky business, and difficult to know if you succeeded, never mind simulating an entire mammalian brain. When you claim to have simulated a whole brain, then you need some kind of holistic results to justify the conclusion. It seemed to me that the results claimed in the original link were rather vague. Seeing catlike patterns doesn't mean much. Was the simulated cat brain able to learn?

Apart from learning, the remarkable things about a cat brain need a cat body to show off. If you have ever watched a cat walk along the top of a narrow ledge and jump to land perfectly balanced on a higher fence at a ninety degree angle, you have witnessed a miracle of coordination and grace.

Without know the internal details of the supposed simulation of a cat brain, and with no high level results other than similar patterns, I have to remain very skeptical of the achievement.

Title: Re: wow, a neuronal net outgrowing a brain's cat
Post by RonWeasley on Nov 27^th, 2009, 12:48pm

I've had enough cats to question whether simulating their brain is any feat at all. I had one forget where its food was placed. Another can't figure out how to go through a cat door.

I just saw Goyle try a wash spell at the sink. He ended up suddenly covered in dog hair. Somewhere, I suppose, a dog is cleaner, or maybe just wetter. So maybe cats aren't so dumb.

Title: Re: wow, a neuronal net outgrowing a brain's cat
Post by Polyfractal on Dec 16^th, 2009, 6:55pm

I'm on the same boat as Markram on this one. The IBM announcement was completely deceiving. More to the point, their approach is the bog standard computer science to AI. Bigger, faster, stronger will generate human intelligence some day, right? Right?

Markram's point is that they are doing nothing to advance the understanding of intelligence. The brain is more than a collection of point neurons connected together, no matter how many you have. Mohda's simulation bears little resemblance to a real brain, and calling it "cat-scale" brain is a completely worthless comparison.

Edit:

Quote:

It had been my understanding that simulating even one neuron was a tricky business

Depends on the fidelity of simulation that you want. The basic action potential is well characterized. We have precise equations that describe the physics involved. This can describe simple, stereotypical action potential propagating down an axon. The physics in elaborate morphology patterns (like a real neuron) are much more complicated. And the mechanisms that govern when a neuron fires, relative importance of synaptic partners, synaptic pruning and what contributes to changes in plasticity are still largely up for debate.

As an aside, I'm a neuroscience major if anyone has any questions about this stuff :)