Engineers at Columbia University in New York have used artificial intelligence (AI) to discover that fingerprints from different fingers of the same person are not unique, shattering an old belief in forensic science, the institution said this Wednesday.
The work, published in Science Advances, a journal of the American Association for the Advancement of Science (AAAS), shows that these “intrapersonal fingerprints (…) are similar,” but their comparison was done “in the wrong way.” .
“It’s not just detectives from the American TV series Law & Order or forensic scientists from crime scene investigators CSI who use fingerprints as a template to link criminals to a crime. In real life, researchers also do the same. “, Columbia University’s Faculty of Engineering and Applied Sciences said in a statement.
But if a criminal leaves different fingerprints at two crime scenes, it is very difficult to link them, he adds.
The team that carried out the work was led by Gabe Guo, a senior engineering student at Columbia University with no prior knowledge of forensics.
Guo found a publicly available database containing about 60,000 fingerprints and “entered them in pairs into an artificial intelligence-based system known as a deep contrast network.” The pairs could belong to the same person (but with different fingers) or to different people.
Over time, the artificial intelligence system the team developed by modifying the “state-of-the-art framework” improved its ability to detect fingerprints that belonged to the same person and those that did not.
“The accuracy for one pair reached 77%. When multiple pairs were used, the degree of accuracy was significantly higher, potentially increasing forensic efficiency by more than tenfold.”
On the other hand, Guo realized that the AI ”was not using the patterns used in traditional fingerprint comparison,” the “minutiae” (“branches and ends of ridges”), but rather something “related to the angles and curvature of fingerprints.” spirals at the center of digital printing.”
Disclosure of the project – a collaboration between the Creative Machines laboratories led by engineer Hod Lipson and Embedded Sensors and Computing led by Wenyao Xu, a professor at the State University of New York at Buffalo – has been rejected more than once, the first by a forensic audit.
Lipson appealed the second denial, deeming “the discovery too important to ignore.”
Although the system’s accuracy is not sufficient to formally resolve a case, it can help prioritize ambiguous situations.
“If this information tips the scales, I think cold cases could be reopened and even innocent people could be exonerated,” Lipson said, as quoted in the statement.
While the study authors presented evidence that the AI system “performed similarly across genders and races for which samples were available,” they are aware of possible bias in the data.
Therefore, they believe that “more thorough validation using datasets with wider coverage” is needed before this method is used in practice.
Author: Lusa
Source: CM Jornal
I’m Tifany Hawkins, a professional journalist with years of experience in news reporting. I currently work for a prominent news website and write articles for 24NewsReporters as an author. My primary focus is on economy-related stories, though I am also experienced in several other areas of journalism.