mlsanityData to 7 Oct 2026

Mind the Graph When Balancing Data for Fairness or Robustness

arXiv 2406.17433 ↗First posted 25 Jun 2024cs.LGcited by 9

Not on the map. Why?

Abstract (arXiv)

Failures of fairness or robustness in machine learning predictive settings can be due to undesired dependencies between covariates, outcomes and auxiliary factors of variation. A common strategy to mitigate these failures is data balancing, which attempts to remove those undesired dependencies. In this work, we define conditions on the training distribution for data balancing to lead to fair or robust models. Our results display that, in many cases, the balanced distribution does not correspond to selectively removing the undesired dependencies in a causal graph of the task, leading to multiple failure modes and even interference with other mitigation techniques such as regularization. Overall, our results highlight the importance of taking the causal graph into account before performing data balancing.

Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 44 of its 82 references that are in the map. First posted on arXiv 25 Jun 2024.

Builds on

44 of 82 references.

References · 44

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  3. 2020Training data-efficient image transformers & distillation through attentioncited by 4,560
  4. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  5. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  6. 2016Equality of Opportunity in Supervised Learningcited by 2,000
  7. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,911
  8. 2019Invariant Risk Minimizationcited by 1,832
  9. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  10. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 379
  11. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  12. 2018What is the Effect of Importance Weighting in Deep Learning?cited by 250
32 more
  1. 2016Statistics of Robust Optimization: A Generalized Empirical Likelihood Approachcited by 182
  2. 2018Path-Specific Counterfactual Fairnesscited by 179
  3. 2021Simple data balancing achieves competitive worst-group-accuracycited by 177
  4. 2020Characterising Bias in Compressed Modelscited by 162
  5. 2021Reading Race: AI Recognises Patient's Racial Identity In Medical Imagescited by 155
  6. 2019Identifying and Correcting Label Bias in Machine Learningcited by 146
  7. 2017Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributionscited by 125
  8. 2023Discover and Cure: Concept-aware Mitigation of Spurious Correlationcited by 71
  9. 2021Causally motivated Shortcut Removal Using Auxiliary Labelscited by 55
  10. 2023Towards Last-layer Retraining for Group Robustness with Fewer Annotationscited by 54
  11. 2022Detecting Shortcut Learning for Fair Medical AI using Shortcut Testingcited by 52
  12. 2021A Systematic Review of Robustness in Deep Learning for Computer Vision: Mind the gap?cited by 48
  13. 2023Mitigating Spurious Correlations in Multi-modal Models during Fine-tuningcited by 47
  14. 2020Fair Bayesian Optimizationcited by 45
  15. 2022Diagnosing failures of fairness transfer across distribution shift in real-world medical settingscited by 45
  16. 2021Out-of-distribution Generalization in the Presence of Nuisance-Induced Spurious Correlationscited by 41
  17. 2018Fair and Diverse DPP-based Data Summarizationcited by 39
  18. 2024CLIP the Bias: How Useful is Balancing Data in Multimodal Learning?cited by 35
  19. 2022Modeling the Data-Generating Process is Necessary for Out-of-Distribution Generalizationcited by 34
  20. 2023Overwriting Pretrained Bias with Finetuning Datacited by 32
  21. 2021Representation Matters: Assessing the Importance of Subgroup Allocations in Training Datacited by 31
  22. 2022Are All Spurious Features in Natural Language Alike? An Analysis through a Causal Lenscited by 28
  23. 2020Causal Feature Selection for Algorithmic Fairnesscited by 25
  24. 2023Fairly Adaptive Negative Sampling for Recommendationscited by 22
  25. 2021A Near-Optimal Algorithm for Debiasing Trained Machine Learning Modelscited by 19
  26. 2023When More is Less: Incorporating Additional Datasets Can Hurt Performance By Introducing Spurious Correlationscited by 18
  27. 2023Last-Layer Fairness Fine-tuning is Simple and Effective for Neural Networkscited by 17
  28. 2023Causal Context Connects Counterfactual Fairness to Robust Prediction and Group Fairnesscited by 17
  29. 2023Improving Fair Training under Correlation Shiftscited by 14
  30. 2022Deep Learning Discovery of Demographic Biomarkers in Echocardiographycited by 12
  31. 2023Spurious Correlations and Where to Find Themcited by 6
  32. 2023The Limits of Fair Medical Imaging AI In The Wildcited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 5; 2026: 3.

9 papers.

  1. 2025Person-Centric Annotations of LAION-400M: Auditing Bias and Its Transfer to Modelscited by 4
  2. 2025Subgroups Matter for Robust Bias Mitigationcited by 1
  3. 2025Does Weak-to-strong Generalization Happen under Spurious Correlations?cited by 1
  4. 2026Benchmarking Bias Mitigation Toward Fairness Without Harm from Vision to LVLMscited by 1
  5. 2024Causal Fine-Tuning under Latent Confounded Shiftcited by 0
  6. 2025A Critical Review of Predominant Bias in Neural Networkscited by 0
  7. 2025Representation Invariance and Allocation: When Subgroup Balance Matterscited by 0
  8. 2026DeconDTN-Toolkit: A Library for Evaluation and Enhancement of Robustness to Provenance Shiftcited by 0
  9. 2026Structure-agnostic Causal Representation Learningcited by 0