Engineers at the universities of Columbia and Buffalo have developed a new fingerprint analysis using artificial intelligence (AI). This analysis challenges the idea that no two fingerprints are ever the same, not even on different fingers of the same person.
The study, published in the journal Science Advance, demonstrates with 99.99% reliability that the fingerprints of any two fingers of the same person are much more similar than previously believed.
Fingerprints are key pieces in criminalistics and digital authentication on millions of mobile devices. However, so far, the existing technology is based on the premise of the uniqueness of each fingerprint.
The research group, led by Gabe Guo, an engineering student at Columbia, along with other researchers from the same university and the University of Buffalo, used a public US government database with about 60,000 fingerprints and applied a system of AI known as deep contrast network.
An unexpected find
Using deep neural networks, they analyzed 525,000 fingerprint images, discovering an extraordinary similarity in the prints of different fingers of the same person. This finding came as a surprise, especially to a team of engineers with no prior forensic experience.
The study revealed that the orientation of the finger ridges, especially near the center, is a determining factor in the observed similarity. This pattern remains constant in all pairs of a person’s fingers, regardless of their gender or racial group.
“We hope this information can help prioritize leads in investigations, exonerate innocent suspects, and create new avenues in unsolved cases,” Guo says. Additionally, the discovery could have a significant impact on improving digital authentication techniques.
With information from DW