Racial Faces in-the-Wild: Reducing Racial Bias by Information Maximization Adaptation Network
This addresses racial bias in biometric systems, which is a critical fairness issue for diverse populations, but the method is incremental as it builds on existing domain adaptation techniques.
The paper tackles racial bias in deep face recognition by introducing the RFW dataset to validate bias in commercial APIs and SOTA algorithms, and proposes the IMAN method to reduce bias, showing improved generalization across races and databases in experiments.
Racial bias is an important issue in biometric, but has not been thoroughly studied in deep face recognition. In this paper, we first contribute a dedicated dataset called Racial Faces in-the-Wild (RFW) database, on which we firmly validated the racial bias of four commercial APIs and four state-of-the-art (SOTA) algorithms. Then, we further present the solution using deep unsupervised domain adaptation and propose a deep information maximization adaptation network (IMAN) to alleviate this bias by using Caucasian as source domain and other races as target domains. This unsupervised method simultaneously aligns global distribution to decrease race gap at domain-level, and learns the discriminative target representations at cluster level. A novel mutual information loss is proposed to further enhance the discriminative ability of network output without label information. Extensive experiments on RFW, GBU, and IJB-A databases show that IMAN successfully learns features that generalize well across different races and across different databases.