IBM researchers have developed a new kind of computer chip that tries to mimic the way brains recognize patterns, relying on densely interconnected webs of transistors similar to the brain's neural network. Named TrueNorth, the processor has electronic "neurons" that are able to signal others when a type of data — such as light — passes a certain threshold. Working in parallel, the neurons begin to organize the data into patterns suggesting the light is growing brighter, or changing color or shape.
The chip contains 5.4 billion transistors but draws just 70 milliwatts of power. The vast number of circuits working in parallel enables the chip to perform 46 billion operations a second per watt of energy consumed. TrueNorth has 1 million neurons, which is about as complex as the brain of a bee. "It is a remarkable achievement in terms of scalability and low power consumption," says the Lawrence Berkeley National Laboratory's Horst Simon.
IBM's work in neuromorphic computing was funded by the U.S. Defense Advanced Research Projects Agency, but neural network expert Yann LeCun doubts IBM's TrueNorth approach will ever outpace the fastest current commercial processors. "This particular task won't impress anyone in computer vision or machine learning," he contends.
From The New York Times
View Full Article – May Require Registration
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found