Learning to Attack the Cyberattackers Can’t Happen Fast Enough

PITTSBURGH — In a technology lab full of graduate students huddled over laptops, Prof. Marios Savvides flipped through photos on a computer screen searching for one full of people whose faces were barely recognizable to the human eye.

“How about a riot?” Professor Savvides asked. He had just come upon an image of police officers wearing helmets and gas masks and rioters covering their mouths and noses with bandannas — all trying to shield themselves from the tear-gas- and smoke-filled air.

Professor Savvides was delighted. It was a perfect example of where, with the facial recognition skills of artificial intelligence, “we can now recognize a face from very few pixels,” he said.

The episode was unfolding at the Biometrics Center, part of the CyLab Security and Privacy Institute at Carnegie Mellon University.

The center was created by Professor Savvides, who is a widely recognized expert in biometrics — the science of measuring and identifying people using facial and iris recognition systems. On any given day, the high-tech space is crowded with computers, robots, and other machines and populated with doctoral students working with him.

CyLab, which includes the center, was founded in 2003 to expand the boundaries of technology and protect people when that technology — or the people using it — poses a threat.