New York, USA – Fingerprint analysis, a tool long relied upon in crime-solving, is being challenged by a team of researchers who have discovered surprising connections between prints from different fingers of the same individual. This team, led by Hod Lipson from Columbia Engineering and in collaboration with Wenyao Xu from the University at Buffalo, used artificial intelligence to break down traditional forensic norms.
Contrary to the widely accepted belief that fingerprints from different fingers of one person do not match, the researchers found that prints can sometimes appear more alike. Through the use of AI technology, the team fed pairs of fingerprints into a deep contrastive network, achieving an accuracy of 77% for single pairs and even higher accuracy for grouped samples.
Despite facing resistance during peer review, with their project initially rejected by a well-established forensics journal, the researchers persisted. Eventually, their study was recognized and published in the peer-reviewed journal, Science Advances, offering new possibilities for connecting crime scenes and potentially reviving cold cases.
Traditional methods in fingerprint analysis focus on minutiae, but the AI used in this study focused on angles and curvatures of swirls and loops in fingerprints. This approach, not reliant on traditional minutiae patterns, may offer new insights for investigators and prompt a reevaluation of current forensic procedures.
Looking ahead, the researchers acknowledge the need for larger, more diverse fingerprint collections to avoid potential biases in their AI system. While the AI cannot conclusively solve legal matters, it can serve as a supplementary tool for law enforcement to narrow down suspects or connect different crime scenes based on partial matches.
The study’s findings underscore the potential of AI technology to uncover patterns that traditional methods may overlook, emphasizing the importance of open datasets in research. With the increasing role of AI in scientific discovery, the researchers anticipate a shift in how experts approach forensic investigations, welcoming unexpected breakthroughs from fresh perspectives.
As technology continues to advance, the researchers believe that AI-led discoveries by non-experts will become more prevalent, shaping the future of forensic analysis. This study marks a significant step in challenging long-held beliefs in the field of fingerprint analysis, paving the way for innovative approaches to crime-solving in the years to come.