The key component of the computer "brain" under development by IBM and the government is the modeling of synapse connections between neurons.  (Source: Science Photo Library)
IBM is trying to develop circuits that mimic the human brain

Think that computer upgrades could one day make the human brain obsolete?  You're not alone.  However, to reach that critical milestone key upgrades in computing will be needed to make computers more brain-like in operation.

To those ends IBM is taking the lead on a major government research endeavor in the field of "cognitive computing" which pairs neurobiologists, computer and materials scientists and psychologists in a $4.9M USD DARPA-grant driven project to develop a computer that behaves like a brain, down to the neuron level.

Dharmendra Modha, the IBM scientist who is heading the collaboration describes, "The mind has an amazing ability to integrate ambiguous information across the senses, and it can effortlessly create the categories of time, space, object, and interrelationship from the sensory data.  There are no computers that can even remotely approach the remarkable feats the mind performs.  The key idea of cognitive computing is to engineer mind-like intelligent machines by reverse engineering the structure, dynamics, function and behavior of the brain."

The project will utilize an IBM supercomputer as its hardware, a field where IBM has long been king of the hill.  Five universities will devote their talents to making this computer behave like a collection of neurons.  The goal is replicate behavior in simulations.  The long term goal is to create a "brain" on the intelligence level of a cat.

The work will draw heavily from neuroscience, which has mapped out simple animal brains and how they respond to stimuli.  Project leader Mr. Modha has some brain-simulating experience of his own -- last year he led a team which used an IBM BlueGene supercomputer to simulate a mouse brain with 55m neurons and some half a trillion synapses.  He describes, "But the real challenge is then to manifest what will be learned from future simulations into real electronic devices - nanotechnology."

Today electronics can be manufactured at as high a density as animal neurons.

The new effort differs from efforts to establish so-called neural networks.  Neural networks, which seek to simulate connections of neurons and can approach learning-like behavior, and artificial intelligence are inherently different from the attempt to create a full brain.  Says Mr. Modha, "The issue with neural networks and artificial intelligence is that they seek to engineer limited cognitive functionalities one at a time. They start with an objective and devise an algorithm to achieve it.  We are attempting a 180 degree shift in perspective: seeking an algorithm first, problems second. We are investigating core micro- and macro-circuits of the brain that can be used for a wide variety of functionalities."

The result is more of a synaptic network than a neural one.  The key component to which the brain owes its flexibility is the synapse.  Synapses connect neurons together in the brain and it is these connections that help us think. 

Experts worldwide are intrigued by the project, but fear that the US government is underfunding it.  Still, says Christian Keysers, director of the neuroimaging center at University Medical Centre Groningen, "It's an interesting effort, and modeling computers after the human brain is promising."

"If you mod me down, I will become more insightful than you can possibly imagine." -- Slashdot

Most Popular Articles

Copyright 2018 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki