In this work, we present a framework to measure and mitigate intrinsic biases with respect to protected variables --such as gender-- in visual recognition tasks. We show that trained models significantly amplify the association of target labels with gender beyond what one would expect from biased datasets. Surprisingly, we show that even when datasets are balanced such that each label co-occurs equally with each gender, learned models amplify the association between labels and gender, as much as if data had not been balanced! To mitigate this, we adopt an adversarial approach to remove unwanted features corresponding to protected variables from intermediate representations in a deep neural network -- and provide a detailed analysis of its effectiveness. Experiments on two datasets: the COCO dataset (objects), and the imSitu dataset (actions), show reductions in gender bias amplification while maintaining most of the accuracy of the original models.
Image captioning has made substantial progress with huge supporting image collections sourced from the web. However, recent studies have pointed out that captioning datasets, such as COCO, contain gender bias found in web corpora. As a result, learning models could heavily rely on the learned priors and image context for gender identification, leading to incorrect or even offensive errors. To encourage models to learn correct gender features, we reorganize the COCO dataset and present two new splits COCO-GB V1 and V2 datasets where the train and test sets have different gender-context joint di…
Existing public face datasets are strongly biased toward Caucasian faces, and other races (e.g., Latino) are significantly underrepresented. This can lead to inconsistent model accuracy, limit the applicability of face analytic systems to non-White race groups, and adversely affect research findings based on such skewed data. To mitigate the race bias in these datasets, we construct a novel face image dataset, containing 108,501 images, with an emphasis of balanced race composition in the dataset. We define 7 race groups: White, Black, Indian, East Asian, Southeast Asian, Middle East, and Lati…
Johns Hopkins University（约翰斯·霍普金斯大学）；Univ. of Maryland, College Park（马里兰大学帕克分校）；NIST（美国国家标准与技术研究院）
Face recognition networks generally demonstrate bias with respect to sensitive attributes like gender, skintone etc. For gender and skintone, we observe that the regions of the face that a network attends to vary by the category of an attribute. This might contribute to bias. Building on this intuition, we propose a novel distillation-based approach called Distill and De-bias (D&D) to enforce a network to attend to similar face regions, irrespective of the attribute category. In D&D, we train a teacher network on images from one category of an attribute; e.g. light skintone. Then distilling in…
Johns Hopkins University（约翰斯·霍普金斯大学）；University of Maryland, College Park（马里兰大学学院公园分校）
State-of-the-art deep networks implicitly encode gender information while being trained for face recognition. Gender is often viewed as an important attribute with respect to identifying faces. However, the implicit encoding of gender information in face descriptors has two major issues: (a.) It makes the descriptors susceptible to privacy leakage, i.e. a malicious agent can be trained to predict the face gender from such descriptors. (b.) It appears to contribute to gender bias in face recognition, i.e. we find a significant difference in the recognition accuracy of DCNNs on male and female f…
Beijing Forestry University（北京林业大学）；Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）
It becomes urgent to design effective anti-spoofing algorithms for vulnerable automatic speaker verification systems due to the advancement of high-quality playback devices. Current studies mainly treat anti-spoofing as a binary classification problem between bonafide and spoofed utterances, while lack of indistinguishable samples makes it difficult to train a robust spoofing detector. In this paper, we argue that for anti-spoofing, it needs more attention for indistinguishable samples over easily-classified ones in the modeling process, to make correct discrimination a top priority. Therefore…
Internet search affects people's cognition of the world, so mitigating biases in search results and learning fair models is imperative for social good. We study a unique gender bias in image search in this work: the search images are often gender-imbalanced for gender-neutral natural language queries. We diagnose two typical image search models, the specialized model trained on in-domain datasets and the generalized representation model pre-trained on massive image and text data across the internet. Both models suffer from severe gender bias. Therefore, we introduce two novel debiasing approac…
Facial forgery by deepfakes has raised severe societal concerns. Several solutions have been proposed by the vision community to effectively combat the misinformation on the internet via automated deepfake detection systems. Recent studies have demonstrated that facial analysis-based deep learning models can discriminate based on protected attributes. For the commercial adoption and massive roll-out of the deepfake detection technology, it is vital to evaluate and understand the fairness (the absence of any prejudice or favoritism) of deepfake detectors across demographic variations such as ge…
Tsinghua University（清华大学）；University of California, Los Angeles（加州大学洛杉矶分校）
Advanced machine learning techniques have boosted the performance of natural language processing. Nevertheless, recent studies, e.g., Zhao et al. (2017) show that these techniques inadvertently capture the societal bias hidden in the corpus and further amplify it. However, their analysis is conducted only on models' top predictions. In this paper, we investigate the gender bias amplification issue from the distribution perspective and demonstrate that the bias is amplified in the view of predicted probability distribution over labels. We further propose a bias mitigation approach based on post…
Johns Hopkins University（约翰斯·霍普金斯大学）；University of Maryland, College Park（马里兰大学学院公园分校）
Face recognition networks encode information about sensitive attributes while being trained for identity classification. Such encoding has two major issues: (a) it makes the face representations susceptible to privacy leakage (b) it appears to contribute to bias in face recognition. However, existing bias mitigation approaches generally require end-to-end training and are unable to achieve high verification accuracy. Therefore, we present a descriptor-based adversarial de-biasing approach called `Protected Attribute Suppression System (PASS)'. PASS can be trained on top of descriptors obtained…
As transformer architectures become increasingly prevalent in computer vision, it is critical to understand their fairness implications. We perform the first study of the fairness of transformers applied to computer vision and benchmark several bias mitigation approaches from prior work. We visualize the feature space of the transformer self-attention modules and discover that a significant portion of the bias is encoded in the query matrix. With this knowledge, we propose TADeT, a targeted alignment strategy for debiasing transformers that aims to discover and remove bias primarily from query…
Fairness Without Labels: Pseudo-Balancing for Bias Mitigation in Face Gender Classification
National Institute of Informatics, Japan（日本信息机构国家研究所）；INESC-ID, Instituto Superior Técnico, University of Lisbon, Portugal（里斯本大学技术高等学院、INESC-ID，葡萄牙）；University of Amsterdam, The Netherlands（荷兰阿姆斯特丹大学）；LIMSI, CNRS / Université Paris-Saclay, France（法国CNRS/巴黎萨克雷大学LIMSI）
Published studies have suggested the bias of automated face-based gender classification algorithms across gender-race groups. Specifically, unequal accuracy rates were obtained for women and dark-skinned people. To mitigate the bias of gender classifiers, the vision community has developed several strategies. However, the efficacy of these mitigation strategies is demonstrated for a limited number of races mostly, Caucasian and African-American. Further, these strategies often offer a trade-off between bias and classification accuracy. To further advance the state-of-the-art, we leverage the p…
On the Illusion of Gender Bias in Face Recognition: Explaining the Fairness Issue Through Non-demographic Attributes
Paderborn University（帕德博恩大学）；Technical University of Darmstadt（达姆施塔特技术大学）；University of Notre Dame（诺特丹大学）；Johannes Gutenberg University Mainz（美因茨约翰· Gutenberg大学）
Visa Inc.（Visa公司）；Harvard John A. Paulson School Of Engineering And Applied Sciences（哈佛大学约翰·A·保尔森工程与应用科学学院）；FM: Systems（FM: Systems公司）；Autodesk（Autodesk公司）
Gender classification algorithms have important applications in many domains today such as demographic research, law enforcement, as well as human-computer interaction. Recent research showed that algorithms trained on biased benchmark databases could result in algorithmic bias. However, to date, little research has been carried out on gender classification algorithms' bias towards gender minorities subgroups, such as the LGBTQ and the non-binary population, who have distinct characteristics in gender expression. In this paper, we began by conducting surveys on existing benchmark databases for…
University of California at Berkeley（加州大学伯克利分校）；Massachusetts Institute of Technology（麻省理工学院）；California Institute of Technology（加州理工学院）；Amazon Web Services（亚马逊网络服务）
Measuring biases of vision systems with respect to protected attributes like gender and age is critical as these systems gain widespread use in society. However, significant correlations between attributes in benchmark datasets make it difficult to separate algorithmic bias from dataset bias. To mitigate such attribute confounding during bias analysis, we propose a matching approach that selects a subset of images from the full dataset with balanced attribute distributions across protected attributes. Our matching approach first projects real images onto a generative adversarial network (GAN)'…
