A Comprehensive Study on Face Recognition Biases Beyond Demographics
Philipp Terhorst,Jan Niklas Kolf,Marco Huber,Florian Kirchbuchner,Naser Damer,Aythami Morales,Julian Fierrez,Arjan Kuijper +7 more
- pp 1-1
Reads0
Chats0
TLDR
In this paper, the influence of an extended range of facial attributes on the verification performance of two popular face recognition models is investigated, and the results demonstrate that also many non-demographic attributes strongly affect the recognition performance, such as accessories, hair styles and colors, face shapes, or facial anomalies.Abstract:
Face recognition (FR) systems have a growing effect on
critical decision-making processes. Recent works have
shown that FR solutions show strong performance
differences based on the user’s demographics. However, to
enable a trustworthy FR technology, it is essential to know
the influence of an extended range of facial attributes on FR
beyond demographics. Therefore, in this work, we analyse
FR bias over a wide range of attributes. We investigate the
influence of 47 attributes on the verification performance of
two popular FR models. The experiments were performed on
the publicly available MAAD-Face attribute database with
over 120M high-quality attribute annotations. To prevent
misleading statements about biased performances, we
introduced control group based validity values to decide if
unbalanced test data causes the performance differences.
The results demonstrate that also many non-demographic
attributes strongly affect the recognition performance, such
as accessories, hair-styles and colors, face shapes, or facial
anomalies. The observations of this work show the strong
need for further advances in making FR system more robust,
explainable, and fair. Moreover, our findings might help to a
better understanding of how FR networks work, to enhance
the robustness of these networks, and to develop more
generalized bias-mitigating face recognition solutions.read more
Citations
More filters
Journal ArticleDOI
Sensitive loss: Improving accuracy and fairness of face representations with discrimination-aware deep learning
TL;DR: In this paper , a discriminative learning method based on triplet loss function and a sensitive triplet generator is proposed to improve both the accuracy and fairness of biased face recognition algorithms.
Posted Content
MAAD-Face: A Massively Annotated Attribute Dataset for Face Images
Philipp Terhorst,Daniel Fahrmann,Jan Niklas Kolf,Naser Damer,Florian Kirchbuchner,Arjan Kuijper +5 more
TL;DR: The large amount of high-quality annotations from MAAD-Face is made use of to study the viability of soft-biometrics for recognition, providing insights about which attributes support genuine and imposter decisions.
Journal ArticleDOI
Symbolic AI for XAI: Evaluating LFIT Inductive Programming for Explaining Biases in Machine Learning
Alfonso Ortega,Julian Fierrez,Aythami Morales,Zilong Wang,Marina de la Cruz,César Luis Alonso,Tony Ribeiro +6 more
TL;DR: In this paper, the authors use inductive logic programming (ILP) to learn a propositional logic theory equivalent to a given black-box system under certain conditions, and check the viability of learning from interpretation transition (LFIT) in a specific AI application scenario.
Journal ArticleDOI
A survey of automated data augmentation algorithms for deep learning-based image classification tasks
TL;DR: In this article , the authors provide a taxonomy of existing image AutoDA approaches and discuss their pros and cons, and propose several potential directions for future improvements, and identify three key components of a standard AutoDA model: a search space, a search algorithm and an evaluation function.
Journal ArticleDOI
Towards Racially Unbiased Skin Tone Estimation via Scene Disambiguation
TL;DR: In this paper , the authors propose a new evaluation dataset (FAIR) and an algorithm (TRUST) to improve albedo estimation and, hence, fairness, by conditioning on both the face region and a global illumination signal obtained from the scene image.