California State Students Champion Multimodal Biometrics with Security System Development

A team of undergrads at Cal State Fullerton is working on what could be the next big mobile authentication system, writes Jamie Corpuz in the University’s paper, the Daily Titan. The undergraduate researchers, led professors Mikhail Gofman and Sinjini Mitra, is developing a multimodal system that they presented on recently at the CSUF Southern California Conference for Undergraduate Research.

The general aim of the team is to create a more reliable mobile security system using multimodal biometric authentication. More specifically, the group is working on a system that uses short audiovisual recordings to employ facial and speech recognition methods for authentication. So far the team has only been working on the Android platform, but they’re hoping to expand to iOS as their research continues.

The group compared their system’s error rates to those of the Hidden Markov and Fischer Faces algorithms that are currently widely used, and found that while those traditional systems’ standalone use had error rates around 40 percent, the team’s own system had an error rate of only 20 percent. The takeaway is simple: as Professor Mitra puts it, a multimodal system “yields significantly more accurate authentication results than methods that use only one single biometric trait.”

While Apple has been pioneering the widespread use of fingerprint-scanning via its mobile-embedded TouchID system, and Android has been experimenting with its Lollipop OS’ Face Unlock feature, no major mobile device manufacturer has yet employed the kind of multimodal biometric system that the CSUF team is working on, though third party apps are on their way. The Cal State Fullerton team does have their eyes on the prize though: According to Professor Mitra, they “have plans to make our technology available to cell phone manufacturers and other interested people.”