Image source, Getty Images
It is believed that the fingerprints from each finger on a person's hand are completely different. But now this belief is being challenged by research conducted by a Colombian university.
A team from this American university trained an artificial intelligence tool to examine more than 60,000 fingerprints to see if the artificial intelligence could identify which fingerprints came from the same person.
The research team claims that such technology can identify different fingerprints of the same person with up to 75% accuracy, but they are not sure how the fingerprint analysis mechanism works in artificial intelligence.
“We still don't know exactly how this artificial intelligence works,” said Professor Hood Lipson, a robotics expert at Columbia University. Who supervised the study, before
Forensic science
Researchers believe that artificial intelligence analyzes fingerprints in a different way than standard fingerprint analysis methods, as standard fingerprint analysis methods focus on the direction of the raised line in the center of the finger. Not much attention is paid to the termination and branching of convex lines. Also known as points that show important characteristics (details) on the fingerprint.
“It is clearly not using traditional analysis methods that forensic science has used for decades,” Professor Lipson said.
Image source, Getty Images
Standard fingerprint analysis focuses on the direction of the raised line in the middle of the finger.
Professor Lipson said he and Gabe Gu, an undergraduate student, were surprised by the results.
“We are very skeptical of the outcome… We had to check it back and forth several times,” he said.
However, what happened may not be news to other researchers already studying the matter.
Graham Williams, Professor of Forensic Sciences from the University of Hull in the United Kingdom Let's assume that the concept of obtaining unique fingerprints is different. This was not something the industry strongly believed in.
“We actually don't know if fingerprints are unique. The most we can say is that,” he said. As far as we know no two people have ever been shown to have the same fingerprints.
Crime scene
The findings of this study from Columbia University may impact all of forensic science. This includes verifying your identity using biometrics, such as using your fingerprint to unlock electronic devices.
For example, if a thumbprint is discovered at crime scene A. A fingerprint from the index finger was found at the crime scene b. Currently, the forensic process cannot link these two fingerprints to the same person, but an artificial intelligence tool that researchers from Columbia University may be able to link the fingerprints together.
A research team from Columbia University, none of whom had a background in forensic science. Admittedly, more research is still needed.
AI tools are typically trained on massive amounts of data. It is necessary to use a large number of fingerprint databases to develop this technology forward.
In addition, each fingerprint used in the development of this artificial intelligence is a fingerprint that clearly shows the details of the entire finger. While in the real world it is likely that only a small portion of the fingerprint will be found and the details are not entirely clear.
“Our tools are not yet good enough to judge evidence in court. But it is good enough to find evidence in the investigation process,” said Guo, one of the researchers.
But Dr Sarah Fieldhouse, associate professor of forensic science at Staffordshire University, said she did not believe the study would prove its existence. “A major ripple effect” on current criminology work
She said there are still doubts about whether the main features of the fingerprint that AI uses for analysis will remain the same. When the skin of the finger changes shape when the finger comes into contact with the surface or paper used to collect fingerprints. This includes the question of whether AI will continue to evaluate key characteristics from the same fingerprint as a person ages, as traditional standard fingerprint analysis methods take into account fingerprints in the same location even though the owner of your fingerprint will get older.
But this is a difficult question to answer. Because researchers still don't know exactly how this artificial intelligence works. This is no different from the case with many AI tools.
Image source, Getty Images
In the real world, only a small portion of a fingerprint is likely to be found and the details are not entirely clear.
This study from Colombia underwent peer review. (Peer-reviewed) and was published in the journal Science Advances on Friday.
However, there was one pair of twins in Cheshire in northwest England who probably knew this further in advance than anyone else. Their grandmother, Carol, told the BBC that her grandchildren were able to unlock each other's iPhones using their fingerprints.
“They showed it to me on Christmas Day,” she said. “The doctor said my grandchildren were identical twins when they were born. But I can tell the difference between them as they get older.”
She also claimed that her grandchildren can unlock your iPhone by scanning each other's faces as well.
Fingerprints have been identified since before we were born. Research published last year suggests that the genetic process that determines fingerprints may be similar to what happens to animals such as zebras and cheetahs that have different patterns on their bodies, a theory first proposed by Alan Turing in the 1950s.
“Subtly charming student. Pop culture junkie. Creator. Amateur music specialist. Beer fanatic.”