If an alleged perpetrator leaves fingerprints from different fingers at different crime scenes it is very difficult to link the prints to the person, and the case can go cold. The well-accepted gold standard in the forensics community is that fingerprints of different fingers from the same person are unique and therefore unmatchable. But are intra-person fingerprints truly unique?
Despite facing initial skepticism from the forensics community, the researchers were undeterred and enhanced an AI system with data from a public U.S. government database containing approximately 60,000 fingerprints to optimize their results to reveal that fingerprints from different fingers are not unique and unmatchable.
The artificial intelligence-based system known as a deep contrastive network was trained to discern whether pairs of fingerprints (some from the same person but different fingers) belonged to the same person or not. Initially, the system achieved 77% accuracy for single pairs which increased significantly with multiple pairs, which could represent a 10-fold improvement in current forensic efficiency. However, these findings were not well accepted in the forensic community, and they were even rejected by a leading forensics journal that cited the longstanding and well-accepted gold standard belief in fingerprint uniqueness.
“I don’t normally argue editorial decisions, but this finding was too important to ignore,” says Hod Lipson, who is the James and Sally Scapa Professor of Innovation in the Department of Mechanical Engineering and co-director of the Makerspace Facility. “If this information tips the balance, then I imagine that cold cases could be revived, and even that innocent people could be acquitted.”
“The AI was not using ‘minutiae,’ which are the branchings and endpoints in fingerprint ridges — the patterns used in traditional fingerprint comparison,” says Gabe Guo, who began the study as a first-year student at Columbia Engineering in 2021. “Instead, it was using something else, related to the angles and curvatures of the swirls and loops in the center of the fingerprint.”
Although this is just the beginning, the researchers believe that even more improvements can be made in the system’s performance with larger more comprehensive datasets to validate the performance and accuracy across different genders and races.
“Just imagine how well this will perform once it’s trained on millions, instead of thousands of fingerprints,” notes Aniv Ray, senior at Columbia Engineering.
“Many people think that AI cannot really make new discoveries — that it just regurgitates knowledge,” says Lipson. “But this research is an example of how even a fairly simple AI, given a fairly plain dataset that the research community has had lying around for years, can provide insights that have eluded experts for decades.”
“Even more exciting is the fact that an undergraduate student, with no background in forensics whatsoever, can use AI to successfully challenge a widely held belief of an entire field. We are about to experience an explosion of AI-led scientific discovery by non-experts, and the expert community, including academia, needs to get ready,” Lipson continues.
More alarmingly, to play devil’s advocate, if intra-person fingerprints are not unique because we have been looking at them the wrong way, does this hold implications for fingerprints being unique in general? No doubt many people will be following the development of this technology.
Accompanying video: https://www.youtube.com/watch?v=s5esfRbBc18&t=2s