DenseFace: Bias Mitigation in Face Recognition via Density-Aware Probabilistic Matching

Mansur Bultygov, Vadim Selyutin, Dmitry Nekhaev, Ivan Laptev

Keywords: Demographic Bias, Face Recognition, Deep Learning, Statistical Distributions

TL;DR: DenseFace - new method that mitigates demographic bias for any pre-trained face recognition model

Abstract: Despite steady progress in face recognition, current face recognition models still suffer from significant demographic biases. While approaches for bias mitigation have been proposed, existing methods often impose constraints on the training procedure and result in the degradation of recognition accuracy. To address this issue, we here introduce a method that reduces demographic biases of pre-trained face recognition models without compromising their accuracy. To this end, we model face embeddings of each person by von Mises-Fisher (MF) distribution. We next observe the dependency between demographic attributes and the density of MF distributions, and propose DenseFace, a probabilistic face matching procedure that accounts for differences in MF distributions. Our extensive experiments demonstrate DenseFace to consistently reduce biases of strong face recognition models varying in network architectures, training datasets and loss functions. Notably, DenseFace preserves accuracy and requires no retraining of existing face recognition models. Our work also investigates previously adopted bias measures and makes suggestions.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Submission Number: 24343