# Turning a Blind Eye: Explicit Removal of Biases and Variation from Deep Neural Network Embeddings

arXiv: https://arxiv.org/abs/1809.02169 · HTML: https://mlsanity.com/paper/1809.02169/

## Key facts

- Cited by 114 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 6 Sep 2018.

## Built on by (most cited first, 30 of 114)

- [Debiasing Vision-Language Models via Biased Prompts](https://mlsanity.com/paper/2302.00070/) (2023)
- [EnD: Entangling and Disentangling deep representations for bias correction](https://mlsanity.com/paper/2103.02023/) (2021)
- [A Prompt Array Keeps the Bias Away: Debiasing Vision-Language Models with Adversarial Learning](https://mlsanity.com/paper/2203.11933/) (2022)
- [A Comprehensive Study on Face Recognition Biases Beyond Demographics](https://mlsanity.com/paper/2103.01592/) (2021)
- [Fast Model Debias with Machine Unlearning](https://mlsanity.com/paper/2310.12560/) (2023)
- [Audio Self-supervised Learning: A Survey](https://mlsanity.com/paper/2203.01205/) (2022)
- [FairPrune: Achieving Fairness Through Pruning for Dermatological Disease Diagnosis](https://mlsanity.com/paper/2203.02110/) (2022)
- [Leveling Down in Computer Vision: Pareto Inefficiencies in Fair Deep Classifiers](https://mlsanity.com/paper/2203.04913/) (2022)
- [Meta Balanced Network for Fair Face Recognition](https://mlsanity.com/paper/2205.06548/) (2022)
- [Understanding and Mitigating Annotation Bias in Facial Expression Recognition](https://mlsanity.com/paper/2108.08504/) (2021)
- [Discover and Mitigate Unknown Biases with Debiasing Alternate Networks](https://mlsanity.com/paper/2207.10077/) (2022)
- [A Step Toward More Inclusive People Annotations for Fairness](https://mlsanity.com/paper/2105.02317/) (2021)
- [Unbiased Supervised Contrastive Learning](https://mlsanity.com/paper/2211.05568/) (2022)
- [PASS: Protected Attribute Suppression System for Mitigating Bias in Face Recognition](https://mlsanity.com/paper/2108.03764/) (2021)
- [Gender Artifacts in Visual Datasets](https://mlsanity.com/paper/2206.09191/) (2022)
- [Core Knowledge Deficits in Multi-Modal Language Models](https://mlsanity.com/paper/2410.10855/) (2024)
- [Quantifying Societal Bias Amplification in Image Captioning](https://mlsanity.com/paper/2203.15395/) (2022)
- [Measuring Representational Harms in Image Captioning](https://mlsanity.com/paper/2206.07173/) (2022)
- [Perceptual Score: What Data Modalities Does Your Model Perceive?](https://mlsanity.com/paper/2110.14375/) (2021)
- [One Label, One Billion Faces: Usage and Consistency of Racial Categories in Computer Vision](https://mlsanity.com/paper/2102.02320/) (2021)
- [Human-Centric Multimodal Machine Learning: Recent Advances and Testbed on AI-based Recruitment](https://mlsanity.com/paper/2302.10908/) (2023)
- [Learning Bias-Invariant Representation by Cross-Sample Mutual Information Minimization](https://mlsanity.com/paper/2108.05449/) (2021)
- [FairGRAPE: Fairness-aware GRAdient Pruning mEthod for Face Attribute Classification](https://mlsanity.com/paper/2207.10888/) (2022)
- [Survey of Social Bias in Vision-Language Models](https://mlsanity.com/paper/2309.14381/) (2023)
- [Detecting Melanoma Fairly: Skin Tone Detection and Debiasing for Skin Lesion Classification](https://mlsanity.com/paper/2202.02832/) (2022)
- [Efficient fair PCA for fair representation learning](https://mlsanity.com/paper/2302.13319/) (2023)
- [Mitigating Dataset Bias by Using Per-sample Gradient](https://mlsanity.com/paper/2205.15704/) (2022)
- [FairDisCo: Fairer AI in Dermatology via Disentanglement Contrastive Learning](https://mlsanity.com/paper/2208.10013/) (2022)
- [Skin Deep Unlearning: Artefact and Instrument Debiasing in the Context of Melanoma Classification](https://mlsanity.com/paper/2109.09818/) (2021)
- [FLAC: Fairness-Aware Representation Learning by Suppressing Attribute-Class Associations](https://mlsanity.com/paper/2304.14252/) (2023)

## Abstract (arXiv)

Neural networks achieve the state-of-the-art in image classification tasks. However, they can encode spurious variations or biases that may be present in the training data. For example, training an age predictor on a dataset that is not balanced for gender can lead to gender biased predicitons (e.g. wrongly predicting that males are older if only elderly males are in the training set). We present two distinct contributions: 1) An algorithm that can remove multiple sources of variation from the feature representation of a network. We demonstrate that this algorithm can be used to remove biases from the feature representation, and thereby improve classification accuracies, when training networks on extremely biased datasets. 2) An ancestral origin database of 14,000 images of individuals from East Asia, the Indian subcontinent, sub-Saharan Africa, and Western Europe. We demonstrate on this dataset, for a number of facial attribute classification tasks, that we are able to remove racial biases from the network feature representation.