mlsanityData to 13 Oct 2026

Learning from others' mistakes: Avoiding dataset biases without modeling them

arXiv 2012.01300 ↗First posted 2 Dec 2020cs.CL cs.LGcited by 83

Not on the map. Why?

Abstract (arXiv)

State-of-the-art natural language processing (NLP) models often learn to model dataset biases and surface form correlations instead of features that target the intended underlying task. Previous work has demonstrated effective methods to circumvent these issues when knowledge of the bias is available. We consider cases where the bias issues may not be explicitly identified, and show a method for training models that learn to ignore these problematic correlations. Our approach relies on the observation that models with limited capacity primarily learn to exploit biases in the dataset. We can leverage the errors of such limited capacity models to train a more robust model in a product of experts, thus bypassing the need to hand-craft a biased model. We show the effectiveness of this method to retain improvements in out-of-distribution settings even if no particular bias is targeted by the biased model.

Key facts. Cited by 83 papers in the map (data through 13 Oct 2026). First posted on arXiv 2 Dec 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2021: 19; 2022: 24; 2023: 20; 2024: 8; 2025: 10; 2026: 2.

83 papers.

Sort:

By area:

  1. 2021Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasetscited by 181
  2. 2021Evading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions with Superior OOD Generalizationcited by 83
  3. 2022Shortcut Learning of Large Language Models in Natural Language Understandingcited by 83
  4. 2022Distilling Model Failures as Directions in Latent Spacecited by 82
  5. 2021Generalization in NLI: Ways (Not) To Go Beyond Simple Heuristicscited by 67
  6. 2022ZIN: When and How to Learn Invariance Without Environment Partition?cited by 58
  7. 2022Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasetscited by 54
  8. 2021Supervising Model Attention with Human Explanations for Robust Natural Language Inferencecited by 29
  9. 2022SelecMix: Debiased Learning by Contradicting-pair Samplingcited by 29
  10. 2023From Shortcuts to Triggers: Backdoor Defense with Denoised PoEcited by 27
  11. 2023Explanation-based Finetuning Makes Models More Robust to Spurious Cuescited by 25
  12. 2021End-to-End Self-Debiasing Framework for Robust NLU Trainingcited by 24
38 more
  1. 2021Avoiding Inference Heuristics in Few-shot Prompt-based Finetuningcited by 23
  2. 2022BLIND: Bias Removal With No Demographicscited by 22
  3. 2022Causal Balancing for Domain Generalizationcited by 20
  4. 2021Context-aware Adversarial Training for Name Regularity Bias in Named Entity Recognitioncited by 19
  5. 2021RAIL-KD: RAndom Intermediate Layer Mapping for Knowledge Distillationcited by 18
  6. 2021Debiasing Methods in Natural Language Understanding Make Bias More Accessiblecited by 17
  7. 2021Predict then Interpolate: A Simple Algorithm to Learn Stable Classifierscited by 16
  8. 2021Uncertainty Calibration for Ensemble-Based Debiasing Methodscited by 16
  9. 2022Learning to Split for Automatic Bias Detectioncited by 16
  10. 2022Debiasing Masks: A New Framework for Shortcut Mitigation in NLUcited by 16
  11. 2024Causal-Guided Active Learning for Debiasing Large Language Modelscited by 15
  12. 2021Are Bias Mitigation Techniques for Deep Learning Effective?cited by 14
  13. 2021Robustness Challenges in Model Distillation and Pruning for Natural Language Understandingcited by 12
  14. 2022OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypothesescited by 11
  15. 2022Kernel-Whitening: Overcome Dataset Bias with Isotropic Sentence Embeddingcited by 11
  16. 2023Beyond Distribution Shift: Spurious Features Through the Lens of Training Dynamicscited by 11
  17. 2024Analyzing the Evaluation of Cross-Lingual Knowledge Transfer in Multilingual Language Modelscited by 11
  18. 2022AGRO: Adversarial Discovery of Error-prone groups for Robust Optimizationcited by 10
  19. 2022Feature-Level Debiased Natural Language Understandingcited by 10
  20. 2023Robust Natural Language Understanding with Residual Attention Debiasingcited by 10
  21. 2021Combating Unknown Bias with Effective Bias-Conflicting Scoring and Gradient Alignmentcited by 8
  22. 2023Mitigating Bias for Question Answering Models by Tracking Bias Influencecited by 8
  23. 2024Shortcut Learning in In-Context Learning: A Surveycited by 8
  24. 2021A Too-Good-to-be-True Prior to Reduce Shortcut Reliancecited by 7
  25. 2021General Greedy De-bias Learningcited by 7
  26. 2023Distilling Robustness into Natural Language Inference Models with Domain-Targeted Augmentationcited by 7
  27. 2021Learning Stable Classifiers by Transferring Unstable Featurescited by 6
  28. 2022Bridging Fairness and Environmental Sustainability in Natural Language Processingcited by 5
  29. 2023Fighting Bias with Bias: Promoting Model Robustness by Amplifying Dataset Biasescited by 5
  30. 2025EqualizeIR: Mitigating Linguistic Biases in Retrieval Modelscited by 5
  31. 2022Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspectivecited by 4
  32. 2022Robust Task-Oriented Dialogue Generation with Contrastive Pre-training and Adversarial Filteringcited by 4
  33. 2022A Survey on Measuring and Mitigating Reasoning Shortcuts in Machine Reading Comprehensioncited by 4
  34. 2022Less is Better: Recovering Intended-Feature Subspace to Robustify NLU Modelscited by 4
  35. 2023A Simple yet Effective Self-Debiasing Framework for Transformer Modelscited by 4
  36. 2024Conditional Language Learning with Contextcited by 4
  37. 2025Improving the OOD Performance of Closed-Source LLMs on NLI Through Strategic Data Selectioncited by 4
  38. 2021A Generative Approach for Mitigating Structural Biases in Natural Language Inferencecited by 3

50 most cited of 83 All 83 (JSON)