Are fingerprints unique? This is a question that became the focal point of Gabe Guo’s life after a casual chat with a professor during the Covid-19 lockdowns. Little did he know that this conversation would lead to a groundbreaking study challenging the long-accepted truth about fingerprints.
Delving into the Research
Guo, an undergraduate senior in Columbia’s computer science department, spearheaded a study with his coauthor, Professor Wenyao Xu of the University of Buffalo, shedding light on the uniqueness of fingerprints. Despite initial rejections from journals and skepticism from the forensics community, the team persisted, refining their study until the evidence became incontrovertible.
The Unconventional Approach
To achieve their surprising results, the team employed a deep contrastive network, an artificial intelligence model commonly used for facial recognition. Adding a unique twist, they fed the system a database of 60,000 fingerprints, revealing unexpected similarities between fingerprints from different fingers of the same person.
The AI system, with an accuracy peaking at 77% for a single pair, challenged the widely held belief that each fingerprint is entirely unique. Guo explained that the key lies in the angles and curvatures at the center of the fingerprint, providing a rigorous explanation for their findings.
A Shift in Forensic Perspective
Traditionally, forensic analysis has focused on minutiae, the branchings and endpoints in fingerprint ridges. Guo emphasized that while these are excellent for matching fingerprints, they aren’t reliable for finding correlations among fingerprints from the same individual, providing a new insight.
Implications for Criminal Investigations
The authors acknowledge potential biases in their data and stress the need for careful validation across genders and races for real-world forensics use. Despite this, Guo is confident that the discovery can significantly impact criminal investigations. The AI system, by identifying similarities among fingerprints, could generate new leads for cold cases, potentially sparing innocent individuals unnecessary investigations.
Expert Opinions
While some experts like Christophe Champod acknowledge the study’s interest in deep learning techniques on fingerprint images, others, like Simon Cole, downplay its practical utility. Cole argues that the claim that no two fingerprints are exactly alike remains unchallenged, and the study might be overstating its significance.
Beyond Fingerprints
Guo contends that the study’s importance extends beyond fingerprints; it exemplifies the power of AI to automatically recognize and extract relevant features. The open-sourcing of the AI code allows others to scrutinize the results, emphasizing the study as a pioneering step in using AI to unveil hidden patterns in everyday observations.
