US, WASHINGTON (ORDO NEWS) — As the quantum computer race rages between Google and IBM, Intel Inc., a well-known American processor manufacturer and world leader in semiconductors, unveils a neuromorphic computing system capable of emulating the equivalent of 100 million neurons.
What are we talking about Research on neuromorphic computing, an experimental field whose objective, in its processing of information, to operate a computer in the manner of a human brain. It “takes the actual brain process and puts it in silicon,” says Mike Davies, director of the Intel Neuromorphic Computing Laboratory.
With neuromorphic processors, the possibilities for artificial intelligence systems take on another dimension. It is possible to train machine learning models using a fraction of the data commonly used.
As the Wall Street Journal sums it up, this means that these models learn the same way as human babies: by seeing an image only once and by being able to recognize its subject thereafter.
Small mammal
How does it work? The new chip that Intel has just developed is called Pohoiki Springs. It is the direct heir to Pohoiki Beach , a system designed in 2019 by Intel, but it far exceeds its capabilities. Pohoiki Springs comes in the form of a chassis ( “rack” in English) carrying 769 interlaced copies of the Loihi neuromorphic chip, without exceeding 500 watts of energy consumption.
According to the manufacturer, the Loihi chip processes information 1,000 times faster than a conventional processor. Combined in a chassis, the computational capacity of the system reaches 100 million neurons, the equivalent of a small mammal.
Intel also compares its intellectual prowess to that of a mole rat, while Pohoiki Beach was compared to a small insect.
Let’s take some concrete examples. In the case of a general power failure, a neuromorphic system would, for example, be able to immediately identify where energy is most needed, says Mike Davies.
Intel also looked at olfactory computing to test its system. In collaboration with Cornell University, Intel’s neuromorphic computer lab has developed a “digital nose,” trained to mimic the olfactory system of animals.
The Pohoiki Springs system has successfully recognized an odor from a single sample, while a conventional system requires 3,000 olfactory characteristics to achieve a 92% success rate.
During the experiment, the system was able to detect odors such as ammonia or methane, even if these were hidden by other scents.
—
Online:
Our Standards, Terms of Use: Standard Terms And Conditions.
Contact us: [email protected]
The article is written and prepared by our foreign editors from different countries around the world – material edited and published by Ordo News staff in our US newsroom press.