It looks like scientists are one step closer to make AI as efficient as the human brain. Researchers from the MIT-based Center for Brains, Minds, and Machines (CBMM) discovered that the neural network they trained to recognize faces has spontaneously come up with a step that can recognize portraits at different angles.
Training AI to recognize faces
The system is basically a computational model of how the human brains recognize faces, and researchers trained it to identify specific visages by feeding it a set of different orientations for several face templates. In the process of learning to identify faces, the program developed an intermediate processing step, which looked at “a face’s degree of rotation – say 45 degrees from center – but not the direction.”
According to layman’s terms, it means the system that that was looking for similarity between faces was able to do so irrespective of whether a face was tipped over, as long as it was rotated in the same angle. This feature is called “mirror symmetry.”
What excites the researchers most that it copies a previously audited feature of how primates process faces, suggesting that the program might be doing something similar to the human brain. However, they are not quite sure about that.
“This is not a proof that we understand what’s going on,” says Tomaso Poggio, a professor of brain and cognitive sciences at MIT and director of the CBMM. “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.”
Understanding how we recognize faces could be beneficial for the facial recognition systems to become more accurate and significantly better, as they have enormous applications in today’s tech industries. One of them is ‘Face Unlock,’ which is an extremely popular feature in smartphones and laptops. Also, it will let companies like Apple, Google, and Facebook better sort your pictures. It’s still in early stage, but this tiny step towards accomplishing human level facial recognition in computers is a sign that AI is able to replicate some particular functions of human brains.