Getty Images
Fingerprints are no longer unique, and can reopen criminal cases
Science – imagine. Your apartment has just been burgled, your luggage has been turned upside down, your prized possessions have been stolen. A stroke of luck! Forensic police get fingerprints right at your forced door. Thanks to her, a suspect is arrested. However, the culprit’s fingerprints that investigators had were not on the same finger, nor on the same hand…
This scenario seems absurd: fingerprints are unique, not only for each person, but for each finger, including the foot. Unless artificial intelligence is involved. In an experience reported in the magazine Science Advances, a team from Columbia University has shown that, contrary to what was previously believed, our fingerprints are not unique. All fingers of a person have the same “signature”. You just had to know where to look.
Ridges, bumps and curves
All this thanks to artificial intelligence, which is trained to recognize fingerprints in hundreds of real cases, each time from prints coming from the same person or different people. Gradually, something strange happened: the AI managed to identify, with ever-increasing efficiency, which fingerprints came from the same individuals.
” We are talking about identifying the five fingers of the same hand, to match them with the five fingers of the corresponding hand. », clarifies HuffPost Gabriel Guo, one of the researchers who participated in the experiment. Something that humans, so far, have been unable to do. And for good reason: AI doesn’t use the same landmarks as us to recognize fingerprints.
Advanced Science
Conventional techniques for identifying fingerprints, minute by minute.
To confuse a criminal, or to unlock a cell phone, the technique is similar. We rely on the correspondence of characteristic elements, such as the separation of a furrow, or a space that closes – we call it minute. As no fingerprint is identical to another, a certain number of these matches (twelve according to French law) make it possible to formally identify a person.
But our AI doesn’t just see these subtleties, although they’ve always been the element we’ve relied on to identify fingerprints. The algorithms chose to focus on the corners and roundings formed by our fingerprints, especially near the center of the fingers. And from this match, the Columbia computer found with more than 80% accuracy that some of the prints belonged to the same hand or the same person.
Advanced Science
The method used by AI uses roundings in the center of the finger.
Reopen the “cold case”.
Potentially, the impact of this progress is enormous. Enough to use it for police purposes? With this success rate, “ Not for use in court », corrects the young researcher. But by training the AI on a large sample, scientists believe they can achieve perfect efficiency. And in the meantime, this may still make it possible to direct or reopen an investigation.
In fact, how many fingerprints lie dormant in files because they can’t be linked to anyone? Here it is possible to imagine that a properly trained AI would be able to link many cases where different fingerprints, but belong to the same person.
But that’s not all: if the technology proves reliable, why don’t our smartphones and computers rely on these curves to identify their users? Practical when normal unlocking is not hands free. ” This is a simpler app than police identification », Gabriel Guo explains.
To “feed” the AI, the team often used partial prints, which were more or less readable on different backgrounds. In the case of a smartphone that you press on, identification is much easier, and the risk of making a mistake is much less. The interest of this technology has not been lost on industrialists. ” We have received several expressions of interest from several companies », explains the researcher, whose technology is in the process of being patented.
See also on The HuffPost:
