Mitigating Urban-Rural Disparities in Contrastive Representation Learning with Satellite Imagery
Not on the map. Why?
Abstract (arXiv)
Satellite imagery is being leveraged for many societally critical tasks across climate, economics, and public health. Yet, because of heterogeneity in landscapes (e.g. how a road looks in different places), models can show disparate performance across geographic areas. Given the important potential of disparities in algorithmic systems used in societal contexts, here we consider the risk of urban-rural disparities in identification of land-cover features. This is via semantic segmentation (a common computer vision task in which image regions are labelled according to what is being shown) which uses pre-trained image representations generated via contrastive self-supervised learning. We propose fair dense representation with contrastive learning (FairDCL) as a method for de-biasing the multi-level latent space of convolution neural network models. The method improves feature identification by removing spurious model representations which are disparately distributed across urban and rural areas, and is achieved in an unsupervised way by contrastive pre-training. The obtained image representation mitigates downstream urban-rural prediction disparities and outperforms state-of-the-art baselines on real-world satellite images. Embedding space evaluation and ablation studies further demonstrate FairDCL's robustness. As generalizability and robustness in geographic imagery is a nascent topic, our work motivates researchers to consider metrics beyond average accuracy in such applications.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 55 of its 87 references that are in the map. First posted on arXiv 16 Nov 2022.
Builds on
55 of 87 references.
References · 55
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2014Microsoft COCO: Common Objects in Contextcited by 17,498
- 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
- 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
- 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
- 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,845
- 2020Improved Baselines with Momentum Contrastive Learningcited by 2,282
- 2019A Survey on Bias and Fairness in Machine Learningcited by 1,911
- 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,838
- 2018Learning deep representations by mutual information estimation and maximizationcited by 1,265
- 2019Self-Supervised Learning of Pretext-Invariant Representationscited by 806
- 2018Mitigating Unwanted Biases with Adversarial Learningcited by 663
38 more
- 2019Self-supervised Visual Feature Learning with Deep Neural Networks: A Surveycited by 656
- 2020Contrastive Learning with Hard Negative Samplescited by 619
- 2020Dense Contrastive Learning for Self-Supervised Visual Pre-Trainingcited by 512
- 2019Using Self-Supervised Learning Can Improve Model Robustness and Uncertaintycited by 511
- 2020Contrastive learning of global and local features for medical image segmentation with limited annotationscited by 309
- 2021LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentationcited by 259
- 2018Learning Not to Learn: Training Deep Neural Networks with Biased Datacited by 255
- 2019Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigationcited by 252
- 2019Flexibly Fair Representation Learning by Disentanglementcited by 196
- 2020Geography-Aware Self-Supervised Learningcited by 190
- 2019Does Object Recognition Work for Everyone?cited by 149
- 2020Unsupervised Learning of Dense Visual Representationscited by 147
- 2018Invariant Representations without Adversarial Trainingcited by 111
- 2021Understanding and Evaluating Racial Biases in Image Captioningcited by 108
- 2021Are Gender-Neutral Queries Really Gender-Neutral? Mitigating Gender Bias in Image Searchcited by 90
- 2022Self-Supervised Learning of Object Parts for Semantic Segmentationcited by 90
- 2021Self-Supervised Pretraining Improves Self-Supervised Pretrainingcited by 81
- 2022Fair Contrastive Learning for Facial Attribute Classificationcited by 71
- 2020Mitigating Face Recognition Bias via Group Adaptive Classifiercited by 67
- 2021Fair Feature Distillation for Visual Recognitioncited by 59
- 2021Fairness in Cardiac MR Image Analysis: An Investigation of Bias Due to Data Imbalance in Deep Learning Based Segmentationcited by 57
- 2020LoCo: Local Contrastive Representation Learningcited by 55
- 2022Leveling Down in Computer Vision: Pareto Inefficiencies in Fair Deep Classifierscited by 53
- 2020SensitiveLoss: Improving Accuracy and Fairness of Face Representations with Discrimination-Aware Deep Learningcited by 50
- 2020Mitigating Gender Bias in Captioning Systemscited by 50
- 2021Unleashing the Power of Contrastive Self-Supervised Visual Models via Contrast-Regularized Fine-Tuningcited by 50
- 2022Fairness-aware Adversarial Perturbation Towards Bias Mitigation for Deployed Deep Modelscited by 47
- 2021Consistent Instance False Positive Improves Fairness in Face Recognitioncited by 40
- 2019Learning From Brains How to Regularize Machinescited by 37
- 2021Learning Bias-Invariant Representation by Cross-Sample Mutual Information Minimizationcited by 32
- 2020Learning Unbiased Representations via Mutual Information Backpropagationcited by 25
- 2018Gradient Reversal Against Discriminationcited by 20
- 2022A study on the distribution of social biases in self-supervised learning visual modelscited by 13
- 2021Tilted Cross Entropy (TCE): Promoting Fairness in Semantic Segmentationcited by 12
- 2022Segmenting across places: The need for fair transfer learning with satellite imagerycited by 6
- 2021Self-supervised Contrastive Learning for Irrigation Detection in Satellite Imagerycited by 5
- 2016Learning Discriminative Features via Label Consistent Neural Networkcited by 4
- 2022Self-supervised Contrastive Learning for Volcanic Unrest Detectioncited by 4
50 most cited of 55 All 55 (JSON)
Built on by
Citing papers per year: 2023: 1.
1 paper.