(ORDO NEWS) — When artificial intelligence is tasked with visually identifying objects and faces, it allocates certain sections of its network for face recognition – just like the human brain.
The human brain has a dedicated area for face recognition, and the neurons there are so good at what they do that most of us can easily recognize thousands of faces.
Thanks to artificial intelligence, computers can now recognize faces with the same efficiency – and neuroscientists at MIT’s McGovern Institute for Brain Research have found that a computer network trained to recognize faces and other objects uses a surprisingly brain-like strategy for sorting all those objects.
“ The human brain’s solution is to separate the processing of faces from the processing of objects ,” explains Katharina Dobs, one of the authors of the work.
The artificial network she trained did the same. “ And this is the solution that we assume any system trained to recognize faces and classify objects would find ,” she adds.
Dobs collected hundreds of thousands of images to train the neural network to recognize faces and objects.
The collection included over 1,700 faces of various people and hundreds of objects ranging from chairs to burgers. All this was presented to the network without any clues about which one is which.
When the program learned to recognize objects and faces, it organized itself into an information processing network that included blocks specifically designed for face recognition. As in the brain, this specialization appeared in the later stages of image processing.
It is not known how the face processing mechanism arises in the developing brain, but based on their results, the scientists argue that neural networks do not need to have an innate face processing mechanism to acquire this specialization.
—
Online:
Contact us: [email protected]
Our Standards, Terms of Use: Standard Terms And Conditions.