Massachusetts Institute of Technology (MIT) researchers have found the most recent version of deep neural networks match the primate brain.
Because these networks are based on neuroscientists' current understanding of how the brain performs object recognition, the success of the latest networks suggest neuroscientists have a good grasp of how object-recognition works in the human brain, according to MIT professor James DiCarlo. He says this knowledge could lead to better artificial intelligence technologies and new ways to repair visual dysfunction.
First, the researchers measured the brain's object-recognition ability, and then they compared this with representations created by the deep neural networks, which consist of a matrix of numbers produced by each computational element in the system. Each image produces a different array of numbers, and the accuracy of the model is determined by whether it groups similar objects into similar clusters within the representation.
"Through each of these computational transformations, through each of these layers of networks, certain objects or images get closer together, while others get further apart," notes MIT researcher Charles Cadieu.
The researchers now plan to develop models that can mimic other aspects of visual processing, including tracking motion and recognizing three-dimensional forms.
From MIT News
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found