(CNN) — “Do you think every fingerprint is unique?”
That was the question asked by Professor Guo during a casual chat while he was stuck at home during the Covid-19 lockdown waiting for the start of his first year at Columbia University. “Little did I know that conversation would shape the course of my life for the next three years,” Guo said.
Guo, now a senior in the computer science department at Columbia University, led a team that conducted a study on the topic, with Professor Wenyao Xu of the University at Buffalo as one of his co-authors. The study, published this week in the journal Science Advances, appears to cast doubt on a long-accepted fact about fingerprints: According to Guo and his colleagues, not all fingerprints are unique.
In fact, several journals rejected the paper before the team appealed and had it accepted by Science Advances. “At the beginning, there was a lot of opposition from the forensic community,” recalls Guo, who had no experience in the field before this study.
“In the first or second version of our article, they said that it is well known that no two fingerprints are the same. I think this helps a lot in improving our research as we keep putting in more data (and improving accuracy ) until, ultimately, the evidence is indisputable,” he explained.
To achieve the surprising results, the team used an artificial intelligence model called a deep contrast network, which is commonly used for tasks such as facial recognition. The researchers put their own spin on it, feeding it a U.S. government database of 60,000 paired fingerprints, sometimes belonging to the same person (but from different fingers) and sometimes to different people.
In the course of its work, the artificial intelligence-based system discovered that fingerprints from different fingers of the same person have strong similarities and is therefore able to tell when these fingerprints belong to the same person and when they do not, with an accuracy of one pair. 77%, which seems to disprove the claim that every fingerprint is “unique.”
“We found a rigorous explanation: the angle and curvature of the center of the fingerprint,” Guo said.
He added that over hundreds of years of forensic analysis, different features called “signature points,” or branches and endpoints on fingerprint ridges, have been studied as traditional markers for fingerprint identification. “They are great for matching fingerprints, but they are not reliable for finding correlations between fingerprints of the same person,” Guo explained. “That’s what we think.”
The authors stated that they were aware of potential biases in the data. Although they believe the AI system works very similarly across genders and races, according to the study, in order for the system to be useful for real forensics, more careful verification by analyzing larger databases of fingerprints is needed, according to the study.
However, Guo believes the discovery could improve criminal investigations:
“The most immediate application is that it could help find new clues in unsolved cases where fingerprints left at crime scenes come from fingers other than those in the archives,” he said. “But on the other hand, it not only helps It will help catch more criminals. It will also help innocent people so they don’t have to continue being investigated unnecessarily. I think it’s a win for society.”
Christophe Champod, a professor of forensic medicine at the School of Criminal Justice at the University of Lausanne in Switzerland, said that using deep learning technology on fingerprint images is an interesting topic. However, Champaud, who was not involved in the study, said he doesn’t think the work discovered anything new.
“Their argument is that these shapes are somehow related between fingers, and this has been known since the beginning of fingerprint recognition, which was done manually and recorded over many years,” he said. It seems to me that they overestimated their article due to a lack of knowledge. I’m glad they rediscovered something that was already known, but essentially, it’s an exaggeration.”
In this regard, Guo said that no one has quantified or systematically exploited the similarities between different fingerprints of the same person as this new study has done.
“We are the first to clearly state that the similarity is due to the orientation of the central ridge of the fingerprint,” Guo said. “In addition, we are the first to try to match fingerprints from different fingers of the same person, at least using an automated system.”
Simon Cole, a professor in the Department of Criminology, Law and Society at the University of California, Irvine, also thinks the work is interesting but says its practical utility has been overstated. Cole also was not involved in the study.
“We were not ‘wrong’ about the fingerprints,” he said, referring to the forensic experts. “The unsubstantiated but intuitively correct statement that no two fingerprints are ‘identical’ is not refuted by the finding that fingerprints are similar. It is known that fingerprints from different people, as well as fingerprints from the same person, are They are all similar.”
The article said the system could be useful at a crime scene where the fingerprints found were not from fingers in police records, but Cole said that would only happen in special circumstances because when fingerprints are collected , all 10 fingers are collected, and the palm is usually recorded routinely. “It’s unclear to me when they thought law enforcement would only archive some of an individual’s fingerprints, rather than all of them,” he said.
The team that conducted the study was so confident in the results that it open sourced the AI code so that others could test it, a decision that Champod and Cole praised. But Guo said the importance of the study goes beyond fingerprints.
“It’s not just about forensics, it’s about artificial intelligence. Humans have been studying fingerprints for as long as we’ve been around, but no one noticed the similarity until we had artificial intelligence analyze them. This shows how automatically artificial intelligence can The ability to identify and extract relevant fingerprints. Features,” he said.
“I think this study is just the first domino in a series of things. We’re going to see people using artificial intelligence to discover things that are hidden from us, right in front of us, like our fingers.”