Gabriele Fossi
abstract
AI can exhibit bias in different domains such as healthcare [1], recruit-
ing [2], and insurance [3], impacting disadvantaged populations which can
be subject to less accurate predictions. This project focuses on bias in
Face Recognition applications. In particular, how the prediction perfor-
mance of demographic attributes such as age and gender changes across
different ethnicities, since errors and biases in this kind of systems can
have serious consequences. In order to investigate bias, the UTKFace
dataset will be used since besides face images, it contains as well demo-
graphic data: age, gender and ethnicity. This project uses InsightFace, an
open-source library for 2D and 3D deep face analysis, that will be used to
create the embeddings of facial images in the UTKFace dataset. On top
of the embeddings, classification layers will be added in order to be able
to predict age, gender and ethnicity. The model will be trained on a ran-
domly sampled subset of the UTKFace dataset to maintain the original
distribution of the dataset. After training, the accuracy will be computed,
together with additional metrics such as group-wise classification accuracy
and equalized odds, to assess potential biases across different groups. Fi-
nally, the dataset will be balanced to mitigate any existing biases, and the
model will be retrained from scratch to assess its impact on the fairness
and performance of the predictions.
outcomes