AI-based methods for processing human signals in the field of biometrics, soft-biometrics, facial analytics, health care, autonomous driving, and more are very active areas of research and commercialization. However, the deployment of these technologies has unearthed concerns regarding equitable performance. As an example, in the case of biometric systems that are fielded to serve large and diverse population groups, issues of performance differences between individual subgroups (e.g., due to age, gender, and race) have to be systematically addressed. This workshop seeks to understand the current state of knowledge regarding such performance differences, a basic understanding of the sources of differential performance, and mechanisms for their mitigation, especially in the areas of biometrics, soft-biometrics, facial analytics, health, and related technologies.

Topics related to the theme of this workshop continue to draw widespread attention from local, national and international legislators, and worldwide news outlets. This subject is part of an ongoing study on face recognition conducted by the National Academy of Sciences.  As biometric applications are fielded to serve whole populations, concerns about accuracy differences between population sub-groups are increasingly important.  This workshop will seek to understand the current state of knowledge about differences in the major biometric modalities, such as face, iris, fingerprint, and voice, and how such “gaps” might be addressed. 

This half-day workshop will be organized as a single track. The workshop has been designed to engage the audience through a (a) keynote talk, (b) oral/poster presentations by active researchers of peer-reviewed papers, and (c) a panel discussion facilitated by distinguished experts.