mlsanityData to 14 Oct 2026

When majority rules, minority loses: bias amplification of gradient descent

arXiv 2505.13122 ↗First posted 19 May 2025cs.LG cs.AI math.OCcited by 1

Not on the map. Why?

Abstract (arXiv)

Despite growing empirical evidence of bias amplification in machine learning, its theoretical foundations remain poorly understood. We develop a formal framework for majority-minority learning tasks, showing how standard training can favor majority groups and produce stereotypical predictors that neglect minority-specific features. Assuming population and variance imbalance, our analysis reveals three key findings: (i) the close proximity between ``full-data'' and stereotypical predictors, (ii) the dominance of a region where training the entire model tends to merely learn the majority traits, and (iii) a lower bound on the additional training required. Our results are illustrated through experiments in deep learning for tabular and image classification tasks.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 31 of its 62 references that are in the map. First posted on arXiv 19 May 2025.

Builds on

31 of 62 references.

References · 31

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  4. 2016Equality of Opportunity in Supervised Learningcited by 2,007
  5. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  6. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
  7. 2016SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model sizecited by 1,060
  8. 2014Certifying and removing disparate impactcited by 715
  9. 2016Fair prediction with disparate impact: A study of bias in recidivism prediction instrumentscited by 672
  10. 2019TabNet: Attentive Interpretable Tabular Learningcited by 648
  11. 2018Gradient Descent Finds Global Minima of Deep Neural Networkscited by 574
  12. 2017Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraintscited by 420
19 more
  1. 2020The Pitfalls of Simplicity Bias in Neural Networkscited by 351
  2. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  3. 2020Fairness in Machine Learningcited by 194
  4. 2022Bias Mitigation for Machine Learning Classifiers: A Comprehensive Surveycited by 125
  5. 2020Fair Regression with Wasserstein Barycenterscited by 100
  6. 2016The Landscape of Empirical Risk for Non-convex Lossescited by 86
  7. 2024Heavy-Tailed Class Imbalance and Why Adam Outperforms Gradient Descent on Language Modelscited by 80
  8. 2022A Systematic Study of Bias Amplificationcited by 65
  9. 2021Directional Bias Amplificationcited by 54
  10. 2020A survey of bias in Machine Learning through the prism of Statistical Parity for the Adult Data Setcited by 35
  11. 2022Fair learning with Wasserstein barycenters for non-decomposable performance measurescited by 24
  12. 2020Review of Mathematical frameworks for Fairness in Machine Learningcited by 21
  13. 2022Men Also Do Laundry: Multi-Attribute Bias Amplificationcited by 21
  14. 2022A Theoretical Analysis of the Learning Dynamics under Class Imbalancecited by 20
  15. 2020Ridge Regularization: an Essential Concept in Data Sciencecited by 19
  16. 2022Simplicity Bias Leads to Amplified Performance Disparitiescited by 11
  17. 2022Bias-inducing geometries: an exactly solvable data model with fairness implicationscited by 9
  18. 2024An Effective Theory of Bias Amplificationcited by 5
  19. 2023How regularization affects the geometry of loss functionscited by 1

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025Mode Collapse of Mean-Field Variational Inferencecited by 0