Google ‘brain’ recognizes cat images
The project could mark an important advance in the effort to create machines that are able to interact with the world — by recognizing shapes and objects — as effortlessly as humans do.
A Google team connected 16,000 computer processors to create a “neural network” with more than one billion connections.
It turned out that this “Google brain” behaved in a manner that mirrored the habits of many human Internet users: it became obsessed with pictures of cats.
The researchers fed the machine random images plucked from 10 million YouTube videos. Despite none of the images being labelled as feline, the Google brain constructed a kind of ghostly abstract model of a cat after being bombarded by millions of pictures.
“We never told it during the training, ‘This is a cat,’ ” said Jeff Dean, a Google scientist.
“It basically invented the concept of a cat,” Dean added.
By the end of the experiment the Google brain recognized three-quarters of the cats that it was presented with from a collection of 20,000 objects.
A human face, as ’invented’ by Google’s neural network . Telegraph.co.uk
It also proved sensitive to human bodies and faces and performed more than twice as accurately as any previous neural network, according to the New York Times. The researchers who conducted the experiment believed that the process may have mirrored what happens in the human visual cortex, which is the largest system in the human brain and is responsible for processing visual images.
Read the full article at: ottawacitizen.com
"I’m not impressed with your so-called advances.
Let me know when computers can put food in my bowl and then I’ll be impressed."