238
In their study, the scientists used an artificial intelligence model - a deep contrast network that allows for face recognition, writes Success in UA.
The researchers provided the US government with a database containing 60 pairs of fingerprints, which could sometimes belong to the same person (but from different fingers) and sometimes belong to different people.
Professor Guo from Columbia University's Department of Computer Science and colleagues conducted the research. For some time, journals rejected this work, but a team of scientists appealed:
"At first, there was strong resistance from the forensic community. In the first draft or two of our article, they said it was a well-known fact that no two fingerprints are alike. I think it really helped to improve our study because we just kept adding more data to it (increasing the precision) until eventually the evidence was incontrovertible," said Professor Guo.
During observation, the system found that prints from different fingers of the same person are very similar. Thus, the system was able to determine in which cases the fingerprints belonged to the same person with the accuracy of one pair. In this way, scientists refuted the hypothesis that each fingerprint is "unique", it is reported 5.ua.
Professor Guo attributed this fact to the fact that previously, during forensic analysis, attention was paid to the various branches and endpoints of fingerprints, which are commonly used for traditional fingerprint identification markers.
"They are great for fingerprint mapping, but unreliable for detecting correlations between fingerprints of the same person," Guo said.
And while the study's authors said they suspect a potential error in the data, they are confident the discovery will help in future criminal investigations.
Christophe Champod, professor of criminology at the School of Criminal Justice at the University of Lausanne in Switzerland, noted that deep learning methods on fingerprint images are of great interest. At the same time, a scientist who did not participate in the research does not believe that this research was a real discovery.
"Their argument that these shapes correlate to some degree between fingers has been known since the beginning of fingerprinting, when it was done by hand, and has been documented for years," Champod said.
In response, Guo said that no one had ever systematically evaluated and used the similarity between different fingerprints of the same person in the way that the new study did.
"We are the first to directly indicate that the similarity is due to the orientation of the protrusions in the center of the fingerprint. Moreover, we are the first to try to match the fingerprints of different fingerprints of the same person, at least with the help of an automated system," said Guo.
Simon Cole, a professor in the Department of Criminology, Law and Society at the University of California, Irvine, noted the interest of the article and the study itself, but believes that its practical utility and importance are overstated.
"We were not mistaken about the fingerprints. The unproven but intuitively true claim that no two fingerprints are "exactly the same" is not disproved by the fact that the fingerprints are similar. Fingerprints of different people, as well as of the same person, were always similar," said the expert.
In turn, the team of scientists who conducted the research are confident in the results and have opened the source code of the AI for testing by others. Professor Guo himself states that the importance of the study goes far beyond fingerprints:
"It's not just about forensics, it's about artificial intelligence. Humans have been looking at fingerprints for as long as we've existed, but no one ever noticed this similarity until we got our AI to analyze it. It simply speaks to the ability of AI to automatically recognize and obtain the appropriate functions," he said.
