AI study detects that fingerprints can be repeated

During the Covid-19 lockdown, Gabe Guo, a graduate student in the department of computer Science from Columbia University (USA), was challenged by a professor with a question.

The expert intrigued his students by questioning whether fingerprints really are unique to each person. Certainly, we have all reflected on this at least once in our lives.

Three years after the question in the university classroom, Guo is in the last year of the course and plays an important role in defining fingerprints.

In collaboration with Professor Wenyao Xu of the University at Buffalo and other co-authors, Guo conducted research whose results were recently published in the journal Science Advances.

The study, led by Guo, addresses the common notion of uniqueness in fingerprints, leading us to question whether prints really are individual.

Contrary to what everyone thinks about the topic, the study faced challenges during the publication process, being finally accepted after several rejections.

Does everyone really have a unique fingerprint? – Image: Internet/Reproduction

Study reveals new perspectives

The team achieved surprising results by employing an artificial intelligence (AI) model called Deep Contrastive Network, widely used in tasks such as facial recognition.

Adding a unique approach To the model, the researchers fed it with an extensive US government database, made up of 60 thousand fingerprints in pairs, some belonging to the same person (but from different fingers) and others from different people.

The AI-based system revealed strong similarities between the fingerprints of different fingers of the same person, making it possible to accurately distinguish between fingerprints belonging to the same individual and those of other people, achieving a remarkable accuracy of 77% for a single pair.

This finding apparently contradicts the conventional belief that each fingerprint is not repeated.

Over centuries of forensic practice, observation of minutiae such as branches and endpoints on fingerprint ridges has been a traditional method for identification.

However, Gabe Guo challenged this technique. He emphasized that while such details are useful for fingerprint matching, they are not reliable for establishing correlations between fingerprints from the same person.

This perception was crucial in guiding the development of the study that questions the uniqueness of fingerprints.

The researchers revealed that there may be mistakes in the system, even though they believe that the system is completely reliable for such a feat.

Furthermore, Guo revealed that the group of researchers was a pioneer in the detailed analysis of fingerprints, highlighting the importance of continuing to research the topic.