Communications of the ACM
IBM Advances on a Computer that Works Like a Human Brain
Image based on an MRI of the brain then mapped with data provided by the BlueMatter algorithm.
Credit: IBM Almaden; Stanford University
Researchers from IBM's Almaden Research Center and the Lawrence Berkeley National Laboratory have performed a computer simulation that matches the scale and complexity of a cat's brain, while researchers from IBM and Stanford University say they have developed an algorithm for mapping the human brain in unprecedented detail. The researchers say these efforts could help build a computer that replicates the complexity of the human brain.
In the first project, an IBM supercomputer at the Lawrence Livermore Lab was used to model the movement of data through a structure with 1 billion neurons and 10 trillion synapses, enabling researchers to observe how information "percolates" through a system similar to a feline cerebral cortex. The research is part of IBM project manager Dharmendra Modha's efforts to design a new computer by first better understanding how the brain works. "The brain has awe-inspiring capabilities," Modha says. "It can react or interact with complex, real-world environments, in a context-dependent way. And yet it consumes less power than a light bulb and it occupies less space than a two-liter bottle of soda."
Modha says a major difference between the brain and traditional computers is that current computer are designed on a model that differentiates between processing and storing data, which can lead to a lag in updating information. However, the brain can integrate and react to a constant stream of sights, sounds, and sensory information. Modha imagines a cognitive computer capable of analyzing a constant stream of information from global trading floors, banking institutions, and real estate markets to identify key trends and their consequences; or a computer capable of evaluating pollution using real-time sensors from around the world.
From San Jose Mercury News (CA)
View Full Article
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA 
No entries found