Fair Contrastive Learning for Facial Attribute Classification (2022-03-30T00:00:00.000000Z)

TL;DR

This paper analyzes unfairness caused by supervised contrastive learning and proposes a new Fair Supervised Contrastive Loss (FSCL) for fair visual representation learning, which significantly outperforms SupCon and existing state-of-the-art methods in terms of the trade-off between top-l accuracy and fairness.

Abstract

Learning visual representation of high quality is essential for image classification. Recently, a series of contrastive representation learning methods have achieved preeminent success. Particularly, SupCon [18] outperformed the dominant methods based on cross-entropy loss in representation learning. However, we notice that there could be potential ethical risks in supervised contrastive learning. In this paper, we for the first time analyze unfairness caused by supervised contrastive learning and propose a new Fair Supervised Contrastive Loss (FSCL) for fair visual representation learning. Inheriting the philosophy of supervised contrastive learning, it encourages representation of the same class to be closer to each other than that of different classes, while ensuring fairness by penalizing the inclusion of sensitive attribute information in representation. In addition, we introduce a group-wise normalization to diminish the disparities of intra-group compactness and inter-class separability between demographic groups that arouse unfair classification. Through extensive experiments on CelebA and UTK Face, we validate that the proposed method significantly outperforms SupCon and existing state-of-the-art methods in terms of the trade-off between top-l accuracy and fairness. Moreover, our method is robust to the intensity of data bias and effectively works in incomplete supervised settings. Our code is available at https://github.com/sungho-CoolG/FSCL.

Authors

Pilhyeon Lee

4 papers

H. Byun

3 papers

Sungho Park

1 papers

TL;DR

Abstract

Authors

References61 items

Inter-subject Contrastive Learning for Subject Adaptive EEG-based Visual Recognition

Feature Stylization and Domain-aware Contrastive Learning for Domain Generalization

Learning Action Completeness from Points for Weakly-supervised Temporal Action Localization

Fair Feature Distillation for Visual Recognition

Learning Disentangled Representation for Fair Facial Attribute Classification via Fairness-aware Information Alignment

Intra-Class Uncertainty Loss Function For Classification

Fair Attribute Classification through Latent Space De-biasing

FairFaceGAN: Fairness-aware Facial Image-to-Image Translation

Unsupervised Image-to-Image Translation Via Fair Representation of Gender Bias

Distributionally Robust Neural Networks

Supervised Contrastive Learning

Fairness by Learning Orthogonal Disentangled Representations

A Simple Framework for Contrastive Learning of Visual Representations

Self-Supervised Learning of Pretext-Invariant Representations

Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation

Jointly De-Biasing Face Recognition and Demographic Attribute Estimation

Momentum Contrast for Unsupervised Visual Representation Learning

Self-Training With Noisy Student Improves ImageNet Classification

Gaussian Affinity for Max-Margin Class Imbalanced Learning

Randaugment: Practical automated data augmentation with a reduced search space

Age and gender bias in pedestrian detection algorithms

Contrastive Multiview Coding

On the Fairness of Disentangled Representations

Flexibly Fair Representation Learning by Disentanglement

Learning Not to Learn: Training Deep Neural Networks With Biased Data

Learning Controllable Fair Representations

Balanced Datasets Are Not Enough: Estimating and Mitigating Gender Bias in Deep Image Representations

Learning deep representations by mutual information estimation and maximization

Fine-Grained Face Annotation Using Deep Multi-Task CNN

Gradient Reversal against Discrimination: A Fair Neural Network Learning Approach

Achieving Fairness through Adversarial Learning: an Application to Recidivism Prediction

Unsupervised Feature Learning via Non-parametric Instance Discrimination

Learning Adversarially Fair and Transferable Representations

Disentangling by Factorising

Mitigating Unwanted Biases with Adversarial Learning

Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification

Progressive Growing of GANs for Improved Quality, Stability, and Variation

Counterfactual Fairness

Age Progression/Regression by Conditional Adversarial Autoencoder

Range Loss for Deep Face Recognition with Long-Tailed Training Data

From Face Images and Attributes to Attributes

Equality of Opportunity in Supervised Learning

An overview of gradient descent optimization algorithms

Deep Residual Learning for Image Recognition

Deep Learning Face Attributes in the Wild

Cross-Age Reference Coding for Age-Invariant Face Recognition and Retrieval

Very Deep Convolutional Networks for Large-Scale Image Recognition

ImageNet Large Scale Visual Recognition Challenge

Illumination-Aware Age Progression

Hybrid Deep Learning for Face Verification

A Kernel Two-Sample Test

Fairness through awareness

Exploiting Transferable Knowledge for Fairness-Aware Image Classification

Learning from Failure: De-biasing Classifier from Biased Classifier

Fairness gan: Generating datasets with fairness properties using a generative adversarial network

Faceapp apologizes for building a racist ai

Google photos mistakenly labels black people gorillas

Dogs vs. cats

Deep Learning for Face Veriﬁcation

Visualizing Data using t-SNE

Target classes are biased to sensitive classes with a one-to-one mapping

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names