And that, according to the researchers, is exactly what the AI did, identifying whether prints from different types of fingers came from the same person with 75 to 90 percent accuracy, the BBC reported.
“It is clear that it isn’t using traditional markers that forensics have been using for decades,” study co-author Hod Lipson, a roboticist at Columbia University, told the broadcaster.
The researchers trained their AI model on a database of 60,000 fingerprints. Lead author Gabe Guo, a senior undergrad at Columbia University, told CNN that the AI was able to look beyond finger features known as “minutiae” that detectives have relied on for centuries.
Leave a reply