AI Facial Recognition System Evaluated by BSI: Corsight Achieves First Completion
Catching Bias in Biometrics: iBeta Amps Up Diversity in Testing
In a groundbreaking move, iBeta Quality Assurance has rolled out a new service to combat demographic bias in biometric systems. This innovative solution harps on the ISO/IEC 19795-10 standard, focusing on measuring the variation in a biometric system's performance across different demographics [1]. The final draft of this standard hit the market in 2024 [1].
The new testing methodology emphasizes widening the demographic spectrum of subjects, encompassing segments like age, gender, and skin tone. Under this approach, the Monk scale comes into play for skin tone categorization. After the biometric data is gathered, it undergoes rigorous testing, and the results are subsequently analyzed based on the respective demographic groups. To maintain fairness, normalization techniques are employed to account for disparities in group sample sizes. This meticulous process allows iBeta to assess whether coverage intervals for each demographic group are well-balanced. In essence, it offers a crystal-clear bias analysis for the solutions under scrutiny [1].
iBeta has garnered accreditation to test against the ISO/IEC 19795-10 standard during its biannual audit under NIST’s National Voluntary Laboratory Accreditation Program (NVLAP). iBeta's upper echelon vehemently advocates for the escalating adoption of biometric technology in both governmental and commercial arenas. With this service, they are poised to ensure these technologies cater to the needs of all individuals irrespective of their background. By providing clear metrics and best practices, the new standard and testing service empower providers to prove the fairness and equitability of their solutions [1].
iBeta continues to hold its reputation as a globally admired independent testing lab. Over the years, it has secured accreditation from prestigious organizations, such as the DEA, Mastercard, Google/Android, and the FIDO Alliance [1][3].
The new testing methodology implemented by iBeta incorporates artificial-intelligence, specifically the Monk scale for skin tone categorization, in its efforts to combat bias in biometrics technology. This service, which emphasizes widening the demographic spectrum, is designed to ensure that biometric solutions cater to the needs of all individuals, regardless of their background or demographic.
