FairGRAPE: Fairness-Aware GRAdient Pruning mEthod for Face Attribute Classification

Lin, Xiaofeng; Seungbae, Kim,; Joo, Jungseock

doi:10.1007/978-3-031-19778-9_24

Cited by 12 publications

(7 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, SOS-SE is competitive with recent approaches (e.g. FairGRAPE [11] achieving 90.90%) involving heavy transformer architectures (such as [12] 91.93%). It is also better than baselines CNN8-MT and CNN18-MT (whose architecture mimics VGG19) models trained from scratch on CelebA.…”

Section: B Large-scale Experiments: Application To Facial Attributes ...mentioning

confidence: 94%

Adversarial Deep Multi-Task Learning Using Semantically Orthogonal Spaces and Application to Facial Attributes Prediction

Dapogny

Gauthier²,

Bonnard³

et al. 2023

2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)

View full text Add to dashboard Cite

Deep learning-based multi-task approaches usually rely on factorizing representation layers up to a certain point, where the network splits into several heads, each one addressing a specific task. Depending on the inter-task correlation, such naive model may or may not allow the tasks to benefit from each others. In this paper, we propose a novel Semantic Orthogonality Spaces (SOS) method for multi-task problems, where each task is predicted using the information from a common subspace that factorizes information among all tasks, as well as a task-specific subspace. We enforce orthogonality between these tasks by applying soft orthogonality constraints, as well as adversarially-learned semantic orthogonality objectives that ensures that predicting one task requires the specific information related to that task. We demonstrate the effectiveness of SOS on synthetic data, as well as for large-scale facial attributes prediction. In particular, we use SOS to craft a lightweight architecture that provides high-end accuracies on CelebA database.

show abstract

Section: B Large-scale Experiments: Application To Facial Attributes ...mentioning

confidence: 94%

Adversarial Deep Multi-Task Learning Using Semantically Orthogonal Spaces and Application to Facial Attributes Prediction

Dapogny

Gauthier²,

Bonnard³

et al. 2023

2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)

View full text Add to dashboard Cite

show abstract

“…We classify whether the dermatological condition in each picture is either benign/non-neoplastic or malignant and we use skin tone as the protected attribute, The dataset provides also annotations for gender and age groups. We adopt the same setup of [83]: gender as sensitive attribute, multi-class race as target.…”

Section: A Datasetsmentioning

confidence: 99%

“…By exploiting the FairFace dataset we can instead study the multi-class target case. Specifically, we considered gender as binary sensitive attribute and multi-class race as target, following the same setup [83]. The obtained results are shown in Table 5 with the EO metrics calculated according to the definition in [91].…”

Section: B Multi-class Sensitive Attributes and Targetmentioning

confidence: 99%

Fairness Meets Cross-Domain Learning: A Benchmark of Models and Metrics

Iurada,

Bucci,

Hospedales

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Deep learning-based recognition systems are deployed at scale for real-world applications that inevitably involve our social life. Although of great support when making complex decisions, they might capture spurious data correlations and leverage sensitive attributes (e.g. age, gender, ethnicity). How to factor out this information while maintaining high performance is a problem with several open questions, many of which are shared with those of the domain adaptation and generalization literature which aims at avoiding visual domain biases. In this work, we propose an in-depth study of the relationship between cross-domain learning (CD) and model fairness, by experimentally evaluating 14 CD approaches together with 3 state-ofthe-art fairness algorithms on 5 datasets of faces and medical images spanning several demographic groups. We consider attribute classification and landmark detection tasks: the latter is introduced here for the first time in the fairness literature, showing how keypoint localization may be affected by sensitive attribute biases. To assess the analyzed methods, we adopt widely used evaluation metrics while also presenting their limits with a detailed review. Moreover, we propose a new Harmonic Fairness (HF) score that can ease unfairness mitigation model comparisons. Overall, our work shows how CD approaches can outperform state-of-the-art fairness algorithms and defines a framework with dataset and metrics as well as a code suite to pave the way for a more systematic analysis of fairness problems in computer vision (Code available at: https://github.com/iurada/fairness_crossdomain).

show abstract

Section: Training the Face Analysis Modelmentioning

confidence: 99%

“…We followed the same training and testing protocols as in [55]. Similar to [55] we reported results on the standard deviation between the performance on the protected groups, referred to as ρ(A). In addition to the accuracy difference metric, this helps us quantitatively access the biases present in the model.…”

Section: Training the Face Analysis Modelmentioning

confidence: 99%

Zero-shot demographically unbiased image generation from an existing biased StyleGAN

Jain,

Dholakia,

Memon

et al. 2023

Preprint

View full text Add to dashboard Cite

<p>Face recognition systems have made significant strides thanks to data-heavy deep learning models, but these models rely on large privacy-sensitive datasets. Recent work in facial analysis and recognition have thus started making use of synthetic datasets generated from GANs and diffusion based generative models. These models, however, lack fairness in terms of demographic representation and can introduce the same biases in the trained downstream tasks. This can have serious societal and security implications. To address this issue, we propose a methodology that generates unbiased data from a biased generative model using an evolutionary algorithm. We show results for StyleGAN2 model trained on the Flicker Faces High Quality dataset to generate data for singular and combinations of demographic attributes such as \textit{Black and Woman}. We generate a large racially balanced dataset of 13.5 million images, and show that it boosts the performance of facial recognition and analysis systems whilst reducing their biases. </p>

show abstract

FairGRAPE: Fairness-Aware GRAdient Pruning mEthod for Face Attribute Classification

Cited by 12 publications

References 38 publications

Adversarial Deep Multi-Task Learning Using Semantically Orthogonal Spaces and Application to Facial Attributes Prediction

Adversarial Deep Multi-Task Learning Using Semantically Orthogonal Spaces and Application to Facial Attributes Prediction

Fairness Meets Cross-Domain Learning: A Benchmark of Models and Metrics

Zero-shot demographically unbiased image generation from an existing biased StyleGAN

Contact Info

Product

Resources

About