mlsanityData to 7 Oct 2026

Simple and Fast Group Robustness by Automatic Feature Reweighting

arXiv 2306.11074 ↗First posted 19 Jun 2023cs.LG stat.MLcited by 68

Not on the map. Why?

Abstract (arXiv)

A major challenge to out-of-distribution generalization is reliance on spurious features -- patterns that are predictive of the class label in the training data distribution, but not causally related to the target. Standard methods for reducing the reliance on spurious features typically assume that we know what the spurious feature is, which is rarely true in the real world. Methods that attempt to alleviate this limitation are complex, hard to tune, and lead to a significant computational overhead compared to standard training. In this paper, we propose Automatic Feature Reweighting (AFR), an extremely simple and fast method for updating the model to reduce the reliance on spurious features. AFR retrains the last layer of a standard ERM-trained base model with a weighted loss that emphasizes the examples where the ERM model predicts poorly, automatically upweighting the minority group without group labels. With this simple procedure, we improve upon the best reported results among competing methods trained without spurious attributes on several vision and natural language classification benchmarks, using only a fraction of their compute.

Key facts. Cited by 68 papers in the map (data through 7 Oct 2026). Builds on 32 of its 46 references that are in the map. First posted on arXiv 19 Jun 2023.

Builds on

32 of 46 references.

References · 32

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  4. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  5. 2017A Broad-Coverage Challenge Corpus for Sentence Understanding through Inferencecited by 2,665
  6. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  7. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  8. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,410
  9. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  10. 2019Class-Balanced Loss Based on Effective Number of Samplescited by 1,239
  11. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  12. 2017ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseasescited by 969
20 more
  1. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  2. 2019Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classificationcited by 396
  3. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 379
  4. 2018Confounding variables can degrade generalization performance of radiological deep learning modelscited by 374
  5. 2020Noise or Signal: The Role of Image Backgrounds in Object Recognitioncited by 342
  6. 2019Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNetcited by 260
  7. 2019Hidden Stratification Causes Clinically Meaningful Failures in Machine Learning for Medical Imagingcited by 215
  8. 2020No Subclass Left Behind: Fine-Grained Robustness in Coarse-Grained Classification Problemscited by 208
  9. 2022Surgical Fine-Tuning Improves Adaptation to Distribution Shiftscited by 205
  10. 2021Simple data balancing achieves competitive worst-group-accuracycited by 177
  11. 2022Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlationscited by 171
  12. 2022On Feature Learning in the Presence of Spurious Correlationscited by 145
  13. 2023Change is Hard: A Closer Look at Subpopulation Shiftcited by 119
  14. 2022Spread Spurious Attribute: Improving Worst-group Accuracy with Spurious Attribute Estimationcited by 88
  15. 2022ZIN: When and How to Learn Invariance Without Environment Partition?cited by 58
  16. 2022A Comprehensive Study of Image Classification Model Sensitivity to Foregrounds, Backgrounds, and Visual Attributescited by 56
  17. 2022Diversify and Disambiguate: Learning From Underspecified Datacited by 55
  18. 2021BARACK: Partially Supervised Group Robustness With Guaranteescited by 25
  19. 2021Robust Representation Learning via Perceptual Similarity Metricscited by 23
  20. 2022Chroma-VAE: Mitigating Shortcut Learning with Generative Classifierscited by 17

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 7; 2024: 23; 2025: 25; 2026: 13.

68 papers.

Sort:

By area:

  1. 2023Towards Last-layer Retraining for Group Robustness with Fewer Annotationscited by 54
  2. 2024The Clever Hans Mirage: A Comprehensive Survey on Spurious Correlations in Machine Learningcited by 39
  3. 2024Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotationscited by 30
  4. 2023Discovering environments with XRMcited by 26
  5. 2023Spurious Feature Diversification Improves Out-of-distribution Generalizationcited by 26
  6. 2025Uncovering Memorization Effect in the Presence of Spurious Correlationscited by 20
  7. 2024Decompose-and-Compose: A Compositional Approach to Mitigating Spurious Correlationcited by 18
  8. 2024The Pitfalls of Memorization: When Memorization Hurts Generalizationcited by 16
  9. 2024Unsupervised Concept Discovery Mitigates Spurious Correlationscited by 11
  10. 2024Controllable Prompt Tuning For Balancing Group Distributional Robustnesscited by 10
  11. 2024Navigating the Shortcut Maze: A Comprehensive Analysis of Shortcut Learning in Text Classification by Language Modelscited by 10
  12. 2025Are We Done with Object-Centric Learning?cited by 10
38 more
  1. 2025Project-Probe-Aggregate: Efficient Fine-Tuning for Group Robustnesscited by 9
  2. 2024Clarify: Improving Model Robustness With Natural Language Correctionscited by 8
  3. 2023Mitigating Simplicity Bias in Deep Learning for Improved OOD Generalization and Robustnesscited by 7
  4. 2025Mitigating Spurious Correlation via Distributionally Robust Learning with Hierarchical Ambiguity Setscited by 7
  5. 2024Robustness to Subpopulation Shift with Domain Label Noise via Regularized Annotation of Domainscited by 6
  6. 2024Dataset Distillation via Knowledge Distillation: Towards Efficient Self-Supervised Pre-Training of Deep Networkscited by 6
  7. 2025Group-robust Sample Reweighting for Subpopulation Shifts via Influence Functionscited by 6
  8. 2025DDB: Diffusion Driven Balancing to Address Spurious Correlationscited by 6
  9. 2024Data Debiasing with Datamodels (D3M): Improving Subgroup Robustness via Data Selectioncited by 5
  10. 2024The Group Robustness is in the Details: Revisiting Finetuning under Spurious Correlationscited by 5
  11. 2024BAdd: Bias Mitigation through Bias Additioncited by 5
  12. 2024Adaptive Group Robust Ensemble Knowledge Distillationcited by 5
  13. 2025Gradient Extrapolation for Debiased Representation Learningcited by 4
  14. 2025ShortcutProbe: Probing Prediction Shortcuts for Learning Robust Modelscited by 4
  15. 2025Diverse Prototypical Ensembles Improve Robustness to Subpopulation Shiftcited by 4
  16. 2025The Impact of Coreset Selection on Spurious Correlations and Group Robustnesscited by 4
  17. 2023Annotation-Free Group Robustness via Loss-Based Resamplingcited by 3
  18. 2024Theoretical Guarantees of Data Augmented Last Layer Retraining Methodscited by 3
  19. 2024Not Only the Last-Layer Features for Spurious Correlations: All Layer Deep Feature Reweightingcited by 3
  20. 2024Optimizing importance weighting in the presence of sub-population shiftscited by 3
  21. 2025Improving Group Robustness on Spurious Correlation via Evidential Alignmentcited by 3
  22. 2025On the Unreasonable Effectiveness of Last-layer Retrainingcited by 3
  23. 2023TLDR: Text Based Last-layer Retraining for Debiasing Image Classifierscited by 2
  24. 2024UnLearning from Experience to Avoid Spurious Correlationscited by 2
  25. 2025An XAI-based Analysis of Shortcut Learning in Neural Networkscited by 2
  26. 2025Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergiescited by 2
  27. 2025NeuronTune: Towards Self-Guided Spurious Bias Mitigationcited by 2
  28. 2025Improving Knowledge Distillation Under Unknown Covariate Shift Through Confidence-Guided Data Augmentationcited by 2
  29. 2023Multitask Learning Can Improve Worst-Group Outcomescited by 1
  30. 2024Fighting Spurious Correlations in Text Classification via a Causal Learning Perspectivecited by 1
  31. 2025FairDropout: Using Example-Tied Dropout to Enhance Generalization of Minority Groupscited by 1
  32. 2025Debiased Prompt Tuning in Vision-Language Model without Annotationscited by 1
  33. 2025SAGE: Spuriousness-Aware Guided Prompt Exploration for Mitigating Multimodal Biascited by 1
  34. 2025Let Samples Speak: Mitigating Spurious Correlation by Exploiting the Clusterness of Samplescited by 1
  35. 2026Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglementcited by 1
  36. 2026Automated Background Swapping for Robustness against Spurious Backgroundscited by 1
  37. 2024Mind the GAP: Improving Robustness to Subpopulation Shifts with Group-Aware Priorscited by 0
  38. 2024Feature Protection For Out-of-distribution Generalizationcited by 0

50 most cited of 68 All 68 (JSON)