According to scientists in the field, neuromorphic chips are old news
In last week’s IBM press release, the company revealed its creation of a ¨ new generation of experimental computer chips designed to emulate the brain’s abilities for perception, action and cognition.¨ The two prototype chips are part of IBM’s SyNapse project (Systems of Neuromorphic Adaptive Plastic Scalable Electronics) which employs principles from Neuroscience, Supercomputing and Nanotechnology to form the basis of ¨Cognitive computing¨.
Dharmendra Modha, manager of IBM’s cognitive computing group and head researcher of the SyNAPSE project, describes IBM’s vision of cognitive computing as ¨the quest to engineer mind-like intelligent business machines by reverse engineering the structure, function, dynamics and behavior of the brain….¨
Sound impressive? Absolutely. However, IBM is not the first to claim to be able to build a computer chip akin to the human brain, nor are neuromorphic chips a novel technology. In 1991, Misha A. Mahowald and Carver Mead published a paper in Scientific American titled ¨The Silicon Retina¨ which describes their successful electronic reproduction of the first three of the retina’s five layers. A decade ago, based on Mahold and Mead’s pioneer work, Stanford’s Kareem Zaghloul went on to create an electronic replication of all five layers of the retina which operated on 1,000 times less energy than a PC.
Through the use of nanotechnology, chip fabricators are getting closer to matching the density of the human brain’s cortex by cramming more and more wire and transistors onto smaller squares of silicon. The question is how will immense numbers of transistors be rendered operational on say one square centimetre of silicon?
Does IBM have the answer? Or has there been too much hype? The media sang a similar tune in 2006 when IBM provided some of their researchers to assist in setting up a supercomputer for a project similar to SyNAPSE that was being carried out by the the Swiss government.
Creating computer chips as efficient and powerful as the human brain? The idea may not be completely outlandish but it’s certainly not new.
Read more about neuromorphic chips HERE in an article by Scientific American