Blog...

Say neocortex and most kids will tell you that’s a character from Crash Bandicoot.  But the more practical application of this term refers to the grey matter that surrounds a mammalian brain and is responsible for such functions as language and motor skills, sensory perception, conscious reasoning, and spatial relationships.  Now, a company called Numenta, hopes to tap into the theory behind how the neocortex performs those functions in human thinking and apply them to artificial intelligence.  If successful, the process promises to be able to resolve complex issues at the speed of a super computer.
 
The think tank behind Numenta is headed by three people who are well known in the field of artificial intelligence and technology.  Jeff Hawkins is the Founder.  Hawkins is the lauded pioneer of pattern recognition software that powered his inventions like the Palm Pilot and the Handspring Treo.  Donna Dubinsky, a long time business associate of Hawkins, serves as CEO and co-founder of Numenta.  Finally, rounding out the team, is Dileep George.  George is co-founder and Principle Architect, and is helping to translate Hawkins’ theories into mathematical proof-of-concept.
 
Using Hawkins’ understanding of neocortex functionality, Numenta hopes to create a model of computer programming which can duplicate a human being’s higher reasoning skills.  By doing so, Numenta would be able to offer a multitude of businesses solutions to complicated management issues in a variety of fields.  As BusinessWeek reports, the first test site for Numenta’s product is the ESDA data center.  Data centers take up enormous amounts of energy to their basic functions.  Archiving and retrieving massive amounts of information time and again could be done more efficiently with computer programs that monitor, predict, associate, and direct storage management programs.  “The better companies can analyze their data, the more efficiently they can run energy-hogging data centers,” the BusinessWeek article explains.  While no results have been officially announced, massive data centers like Microsoft and Google should be very interested in Numenta’s progress.
 
While Numenta’s long term goal has yet to be reached, the company has had some limited success with its theories.  In October of 2007, Numenta held a contest called Numenta Challenge: GAME.  The idea was to create a program where players had to perform a task and the result would be judged against the NuPIC Pictures example application.  NuPIC is Numenta’s pattern recognition software which can recognize a hand drawn image against a database of image descriptions and correlate the appropriate image to text association.  The result of the Numenta Challenge yielded some entertaining, and amazing, results.
 
Numenta’s latest efforts to achieve successful replication of neocortex functionality in computers holds a great deal of promise.  Biometric fields could use “deep reasoning” software to analyze a multitude of complex symptoms to arrive at a diagnosis more quickly and more accurately than a human counterpart ever could.  Depending on the level of information available to the program and its ability to associate similar characteristics based off past cases, the results might even be capable of predictive functionality.  But for now, the corporate world must eye Numenta’s baby steps with optimism and nurture its first steps toward a more mature artificial intelligence design.