Combining face biometrics and passwords might improve identity verification
If passwords are of declining use for verifying one’s identity, and face scans look like their effective lifespans will be short as well, how about combining face biometrics with passwords?
An electrical engineering professor at Brigham Young University has filed for a patent for his so-called concurrent two-factor identity verification method of securing data and physical items including devices, vehicles, workplaces and hotel rooms with biometrics.
D.J. Lee’s idea would work like this: a person who record a two-second video of their face while reading a secret series of words. The person also could record themselves mugging briefly — frowning, blinking in a pattern or perhaps even wiggling their ears.
The clip is stored, locally or in the cloud, ready for comparison the next time the person’s identity needs to be verified.
Lee uses an integrated neural network framework that learns biometric face features and actions concurrently. According to BYU, “this framework models dynamic, sequential data … where all the frames in a recording have to be considered.”
Lee and his Ph.D. student Zheng Sun experimented with 8,000 video clips recorded by 50 subjects to train the network, which reached an accuracy of 90 percent. The school says that a larger dataset will push that rate higher.
Exactly matching the original video might be almost impossible, but thresholds would be set to make the process practical.
The goal of the proposed system is to make the identity verification process intentional, greatly increasing the complexity of stealing or manipulating someone’s identity while complicating the user’s life to a far lesser extent.