Scientists Invented New AI Using Micro Eye Movements To Identify People

Scientists have come up with a new deep learning algorithm capable of identifying people by detecting their eye movements, in other words, how they move their eyeballs.

Eye movements, through past cognitive psychology research, has been known to differ from one people to another substantially, just link a fingerprint.

What’s interesting is how these eye movements’ individual characteristics are rather stable and independent of what a person is looking at. That means scientists could use these eye movements as a method of identification.

Researchers from the University of Potsdam, Germany have developed a new method for biometric identification which works the eye’s micro-movements processing. The study, which was published on ArXiv, described the team’s experiment in which they conducted an experiment to study the involuntary eye movements of people and how the findings were used to develop DeepEyedentification. DeepEyedentification is a deep learning architecture which analyzes eye-tracking signals to identify people.

For their study, the scientists looked at 1,000 people’s eyes and discovered that not only their involuntary twitches were unique but they stayed the same. While existing systems required 100 seconds of footage for people identification, the new algorithm can work well with recordings that are only one second long, leading to the technology being much more feasible.

Lena Jäger, a Potsdam scientist, in an interview with Tech Xplore, said: