Brain circuit simulation provides recognition of shapes

A recent paper titled, "Brain Circuit Bottom-Up Engine Simulation and Acceleration using IBM CELL Processor for Vision Applications", makes the claim that two approaches to solving the problem of computer vision, which is a branch of artificial intelligence, lead to their result.

1) a bottom up computer algorithm approach
2) a top down neuron decoding approach

These different approaches lead to an algorithm that can detect lines that make up the edges of a chair. Furthermore, using the sony playstation 3 they can run this algorithm at a much faster speed than would be possible with normal desktop computers. This is because of the many, many sub-units of the IBM Cell processor.

Link:
http://www-304.ibm.com/jct09002c/university/students/contests/cell/r1w1proposal.doc


For information, on the synergistic processing elements (SPE), which make up the "cores", or "subunits" of the IBM Cell: http://en.wikipedia.org/wiki/Cell_microprocessor#Synergistic_Processing_Elements_.28SPE.29