mlsanityData to 7 Oct 2026

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

arXiv 1812.10352 ↗First posted 26 Dec 2018cs.CVcited by 255

Not on the map. Why?

Abstract (arXiv)

We propose a novel regularization algorithm to train deep neural networks, in which data at training time is severely biased. Since a neural network efficiently learns data distribution, a network is likely to learn the bias information to categorize input data. It leads to poor performance at test time, if the bias is, in fact, irrelevant to the categorization. In this paper, we formulate a regularization loss based on mutual information between feature embedding and bias. Based on the idea of minimizing this mutual information, we propose an iterative algorithm to unlearn the bias information. We employ an additional network to predict the bias distribution and train the network adversarially against the feature embedding network. At the end of learning, the bias prediction network is not able to predict the bias not because it is poorly trained, but because the feature embedding network successfully unlearns the bias information. We also demonstrate quantitative and qualitative experimental results which show that our algorithm effectively removes the bias information from feature embedding.

Key facts. Cited by 255 papers in the map (data through 7 Oct 2026). First posted on arXiv 26 Dec 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 53; 2022: 61; 2023: 50; 2024: 49; 2025: 26; 2026: 16.

255 papers.

Sort:

By area:

  1. 2021Domain Generalization: A Surveycited by 760
  2. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 379
  3. 2021Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyondcited by 177
  4. 2021On Calibration and Out-of-domain Generalizationcited by 134
  5. 2021Learning Debiased Representation via Disentangled Feature Augmentationcited by 116
  6. 2022PMR: Prototypical Modal Rebalance for Multimodal Learningcited by 114
  7. 2021Counterfactual Generative Networkscited by 111
  8. 2021EnD: Entangling and Disentangling deep representations for bias correctioncited by 105
  9. 2023Concept-based Explainable Artificial Intelligence: A Surveycited by 97
  10. 2021Evidential Deep Learning for Open Set Action Recognitioncited by 96
  11. 2022Characterizing and overcoming the greedy nature of learning in multi-modal deep neural networkscited by 92
  12. 2021A Survey on Bias in Visual Datasetscited by 84
38 more
  1. 2022Debiasing Graph Neural Networks via Learning Disentangled Causal Substructurecited by 83
  2. 2021Causal Attention for Unbiased Visual Recognitioncited by 80
  3. 2021Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU Modelscited by 78
  4. 2024MMPareto: Boosting Multimodal Learning with Innocent Unimodal Assistancecited by 72
  5. 2022Fair Contrastive Learning for Facial Attribute Classificationcited by 71
  6. 2021BiaSwap: Removing dataset bias with bias-tailored swapping augmentationcited by 70
  7. 2023Fast Model Debias with Machine Unlearningcited by 70
  8. 2022MEDFAIR: Benchmarking Fairness for Medical Imagingcited by 66
  9. 2023A Comprehensive Survey of Forgetting in Deep Learning Beyond Continual Learningcited by 66
  10. 2021Fairness via Representation Neutralizationcited by 63
  11. 2021Continual Learning via Local Module Compositioncited by 61
  12. 2022FairPrune: Achieving Fairness Through Pruning for Dermatological Disease Diagnosiscited by 59
  13. 2022Leveling Down in Computer Vision: Pareto Inefficiencies in Fair Deep Classifierscited by 53
  14. 2022Learning Debiased Classifier with Biased Committeecited by 51
  15. 2022ProMix: Combating Label Noise via Maximizing Clean Sample Utilitycited by 48
  16. 2022Divert More Attention to Vision-Language Trackingcited by 47
  17. 2022Discover and Mitigate Unknown Biases with Debiasing Alternate Networkscited by 47
  18. 2024FairMedFM: Fairness Benchmarking for Medical Imaging Foundation Modelscited by 47
  19. 2021Independent Prototype Propagation for Zero-Shot Compositionalitycited by 42
  20. 2022Optimizing Relevance Maps of Vision Transformers Improves Robustnesscited by 41
  21. 2022Unbiased Supervised Contrastive Learningcited by 40
  22. 2024Core Knowledge Deficits in Multi-Modal Language Modelscited by 38
  23. 2021Perceptual Score: What Data Modalities Does Your Model Perceive?cited by 35
  24. 2022Using Language to Extend to Unseen Domainscited by 34
  25. 2023Measuring axiomatic soundness of counterfactual image modelscited by 34
  26. 2025Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuningcited by 34
  27. 2021Learning Bias-Invariant Representation by Cross-Sample Mutual Information Minimizationcited by 32
  28. 2022Deep Regression Unlearningcited by 32
  29. 2022Avoiding spurious correlations via logit correctioncited by 32
  30. 2023Human-Centric Multimodal Machine Learning: Recent Advances and Testbed on AI-based Recruitmentcited by 32
  31. 2022On Guiding Visual Attention with Language Specificationcited by 30
  32. 2022Bias Mimicking: A Simple Sampling Approach for Bias Mitigationcited by 30
  33. 2022SelecMix: Debiased Learning by Contradicting-pair Samplingcited by 29
  34. 2023Overcoming Simplicity Bias in Deep Networks using a Feature Sievecited by 29
  35. 2022Aligning Eyes between Humans and Deep Neural Network through Interactive Attention Alignmentcited by 27
  36. 2022Sample-Efficient Reinforcement Learning in the Presence of Exogenous Informationcited by 26
  37. 2022How Robust is Unsupervised Representation Learning to Distribution Shift?cited by 24
  38. 2023Model-as-a-Service (MaaS): A Surveycited by 24

50 most cited of 255 All 255 (JSON)