Trust Stamp joins forces with Rwandan Institute to increase equality in biometrics performance

Trust Stamp announced the signing of a Memorandum of Understanding (MoU) with the African Institute for Mathematical Sciences (AIMS) in Kigali, Rwanda to improve biometrics performance.

According to the company, the new agreement underlines a shared desire to “further equity and inclusion for individuals of all races and ethnologies through advancements in biometrics and identity authentication.”

AIMS is one of Africa’s largest networks of Centres of Excellence for innovative post-graduate training in mathematical sciences.

Following the beginning of the partnership, Trust Stamp and AIMS will partner in developing and implementing courses to extend graduate studies programs focused on big data analytics and biometrics.

They will also share academic research, and jointly explore possible national and international funding opportunities to support cooperative agreements.

“This institutional cooperation between Trust Stamp and AIMS marks an important milestone in our long-standing commitment to universal financial inclusion,” comments Trust Stamp CIO Raman Narayanswamy.

The partnership addresses the well-known fact in the industry that certain demographics experience lower authentication accuracy and higher false positive rates (occurring when the biometric system incorrectly matches two faces that are different) with many systems.

High error rates connected to biometric systems’ biases increase the potential of fraud, but they also discriminate against users who may, for instance, not be able to access digital services because of false negatives.

“Empirical evidence shows an ongoing need in the biometrics industry for research and investment into machine learning algorithms with a focus on demographically diverse populations,” Narayanswamy explains.

“Addressing this multifaceted issue is critical to the development of a robust digital infrastructure in Africa while posing wide-reaching implications for equity and inclusion around the world.”

These biased results are sometimes caused by inaccurate algorithms, which in turn are frequently attributed to datasets with disparate proportions of different demographics.

To tackle these issues and create fair face biometrics systems it is necessary to build them using demographically-balanced datasets, which are responsible for training down error rates.

Many researchers are currently working on the issue, with the Idiap Research Institute recently introducing the concept of ‘Fairness Discrepancy Rate’ (FDR) to directly address the issue.

The partnership between Trust Stamp and AIMS will now further these efforts by establishing a collaborative ecosystem between the company and the institution.

“Trust Stamp looks forward to working with AIMS to further identify and mitigate biases in artificial intelligence models that affect the accuracy and performance of biometric technologies across different demographic groups as we continue innovating and delivering accurate, user-centric, and responsible identity solutions that address global needs,” Narayanswamy concludes.

Article Topics

accuracy  |  Africa  |  biometric matching  |  biometric-bias  |  biometrics  |  biometrics research  |  demographic fairness  |  face biometrics  |  Trust Stamp