Communications of the ACM
Advanced Technology May Indicate How Brain Learns Faces
Researchers at the University of Texas at Dallas have found that computer programs called deep convolutional neural networks (DCNNs) figured out how to identify faces in a different way than the researchers expected.
Credit: University of Texas at Dallas
Researchers at the University of Texas at Dallas (UT Dallas) have demonstrated that deep convolutional neural networks (DCNNs) operate similarly to the way human brains do, in terms of identifying faces.
“For the last 30 years, people have presumed that computer-based visual systems get rid of all the image-specific information — angle, lighting, expression and so on,” said Alice O'Toole at UT Dallas.
Previous-generation algorithms were effective in recognizing faces that had only minor changes from images they already knew.
However, current technology knows an identity well enough to recognize faces despite changes in expression, viewpoint, or appearance.
For example, the researchers found the DCNN excelled at connecting caricatures to their corresponding identities.
Said O'Toole, "Given these distorted images with features out of proportion, the network understands that these are the same features that make an identity distinctive and correctly connects the caricature to the identity."
From University of Texas at Dallas News Center
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found