Science

MIT Researchers' AI Mimicked Brain's Facial Recognition

By Christie Abagon , Dec 05, 2016 01:14 AM EST
Close

Are you fascinated by the brain's ability to recognize a face at any angle? MIT scientists may soon be able to answer why.  They were able to design a machine-learning system which is trained to recognize particular faces by feeding it with a battery of sample images.  This breakthrough was revealed at the Center for Brains, Minds, and Machines (CBMM).

MIT's AI And The Brain Are 'Thinking' Along The Same Lines

MIT's unnamed trained system is a model of how the human brain recognizes the face.  It was trained to identify particular visages from a battery of sample images that was fed into it.  The program created intermediate processing step that looked at "a face's degree of rotation - say 45 degrees from center - but not the direction."

Tomaso Poggio, a professor of brain and cognitive sciences at MIT and director of the CBMM, said: "This is not a proof that we understand what's going on.  Models are kind of cartoons of reality, especially in biology. So I would be surprised if things turn out to be this simple. But I think it's strong evidence that we are on the right track."

The Scientists' Machine-Learning System Is An Architecture Of The Human Brain

The system is a neural network which roughly approximates the architecture of the human brain.  Poggio's group trained neural networks to produce representations by memorizing a set of orientations for some faces called, "templates". 

It was not a model that was trying to explain mirror symmetry," Poggio says. "This model was trying to explain invariance, and in the process, there is this other property that pops out."

Christof Koch, president and chief scientific officer at the Allen Institute for Brain Science, said this discovery is a step forward.   "In this day and age, when everything is dominated by either big data or huge computer simulations, this shows you how a principled understanding of learning can explain some puzzling findings," Koch said. 

Related Articles

© 2019 ITECHPOST, All rights reserved. Do not reproduce without permission.
Real Time Analytics