Log In Sign Up

Studying Bias in GANs through the Lens of Race

by   Vongani H. Maluleke, et al.

In this work, we study how the performance and evaluation of generative image models are impacted by the racial composition of their training datasets. By examining and controlling the racial distributions in various training datasets, we are able to observe the impacts of different training distributions on generated image quality and the racial distributions of the generated images. Our results show that the racial compositions of generated images successfully preserve that of the training data. However, we observe that truncation, a technique used to generate higher quality images during inference, exacerbates racial imbalances in the data. Lastly, when examining the relationship between image quality and race, we find that the highest perceived visual quality images of a given race come from a distribution where that race is well-represented, and that annotators consistently prefer generated images of white people over those of Black people.


page 5

page 20

page 21

page 22

page 24

page 27

page 29


FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age

Existing public face datasets are strongly biased toward Caucasian faces...

Digital Image Forensics vs. Image Composition: An Indirect Arms Race

The field of image composition is constantly trying to improve the ways ...

Characterizing patterns in police stops by race in Minneapolis from 2016-2021

The murder of George Floyd centered Minneapolis, Minnesota, in conversat...

Understanding Image Quality and Trust in Peer-to-Peer Marketplaces

As any savvy online shopper knows, second-hand peer-to-peer marketplaces...

Could you become more credible by being White? Assessing Impact of Race on Credibility with Deepfakes

Computer mediated conversations (e.g., videoconferencing) is now the new...

Do I Look Like a Criminal? Examining how Race Presentation Impacts Human Judgement of Recidivism

Understanding how racial information impacts human decision making in on...

When do GANs replicate? On the choice of dataset size

Do GANs replicate training images? Previous studies have shown that GANs...