mlsanityData to 7 Oct 2026

Long-tail learning via logit adjustment

arXiv 2007.07314 ↗First posted 14 Jul 2020cs.LG stat.MLcited by 605

Not on the map. Why?

Abstract (arXiv)

Real-world classification problems typically exhibit an imbalanced or long-tailed label distribution, wherein many labels are associated with only a few samples. This poses a challenge for generalisation on such labels, and also makes na"ive learning biased towards dominant labels. In this paper, we present two simple modifications of standard softmax cross-entropy training to cope with these challenges. Our techniques revisit the classic idea of logit adjustment based on the label frequencies, either applied post-hoc to a trained model, or enforced in the loss during training. Such adjustment encourages a large relative margin between logits of rare versus dominant labels. These techniques unify and generalise several recent proposals in the literature, while possessing firmer statistical grounding and empirical performance.

Key facts. Cited by 605 papers in the map (data through 7 Oct 2026). First posted on arXiv 14 Jul 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 55; 2022: 91; 2023: 125; 2024: 139; 2025: 86; 2026: 109.

605 papers.

Sort:

By area:

  1. 2021Deep Long-Tailed Learning: A Surveycited by 416
  2. 2021Distribution Alignment: A Unified Framework for Long-tail Visual Recognitioncited by 198
  3. 2021CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learningcited by 174
  4. 2022Balanced Contrastive Learning for Long-Tailed Visual Recognitioncited by 143
  5. 2021Contrastive Learning based Hybrid Networks for Long-Tailed Image Classificationcited by 130
  6. 2022Leveraging Real Talking Faces via Self-Supervision for Robust Forgery Detectioncited by 130
  7. 2021Self-supervised Learning is More Robust to Dataset Imbalancecited by 122
  8. 2022Parametric Classification for Generalized Category Discovery: A Baseline Studycited by 113
  9. 2022Federated Learning with Label Distribution Skew via Logits Calibrationcited by 108
  10. 2022Long-Tailed Recognition via Weight Balancingcited by 107
  11. 2022Retrieval Augmented Classification for Long-Tail Visual Recognitioncited by 106
  12. 2021Self-Supervised Aggregation of Diverse Experts for Test-Agnostic Long-Tailed Recognitioncited by 95
38 more
  1. 2022RelTR: Relation Transformer for Scene Graph Generationcited by 95
  2. 2023PromptMRG: Diagnosis-Driven Prompts for Medical Report Generationcited by 92
  3. 2022Balanced MSE for Imbalanced Visual Regressioncited by 89
  4. 2021ACE: Ally Complementary Experts for Solving Long-Tailed Recognition in One-Shotcited by 86
  5. 2022Debiased Learning from Naturally Imbalanced Pseudo-Labelscited by 85
  6. 2022Imbalance Trouble: Revisiting Neural-Collapse Geometrycited by 80
  7. 2021Sample Selection with Uncertainty of Losses for Learning with Noisy Labelscited by 79
  8. 2022Are All Losses Created Equal: A Neural Collapse Perspectivecited by 78
  9. 2022A Survey on Long-Tailed Visual Recognitioncited by 75
  10. 2023Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustmentcited by 75
  11. 2021Label-Imbalanced and Group-Sensitive Classification under Overparameterizationcited by 74
  12. 2021Distilling Virtual Examples for Long-tailed Recognitioncited by 72
  13. 2023DELTA: degradation-free fully test-time adaptationcited by 72
  14. 2022LPT: Long-tailed Prompt Tuning for Image Classificationcited by 69
  15. 2022Nested Collaborative Learning for Long-Tailed Visual Recognitioncited by 63
  16. 2023Decoupled Kullback-Leibler Divergence Losscited by 58
  17. 2023Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurtscited by 58
  18. 2021DASO: Distribution-Aware Semantics-Oriented Pseudo-label for Imbalanced Semi-Supervised Learningcited by 51
  19. 2021Exploring Classification Equilibrium in Long-Tailed Object Detectioncited by 51
  20. 2021Trustworthy Long-Tailed Classificationcited by 51
  21. 2021Adversarial Robustness under Long-Tailed Distributioncited by 50
  22. 2021A Simple Long-Tailed Recognition Baseline via Vision-Language Modelcited by 49
  23. 2022AutoBalance: Optimized Loss Functions for Imbalanced Datacited by 49
  24. 2022ProMix: Combating Label Noise via Maximizing Clean Sample Utilitycited by 48
  25. 2023Synthetic Datasets for Autonomous Driving: A Surveycited by 48
  26. 2023Class-Balancing Diffusion Modelscited by 47
  27. 2023Understanding Imbalanced Semantic Segmentation Through Neural Collapsecited by 46
  28. 2021Structured Sparse R-CNN for Direct Scene Graph Generationcited by 44
  29. 2022Partial and Asymmetric Contrastive Learning for Out-of-Distribution Detection in Long-Tailed Recognitioncited by 44
  30. 2024CoRAL: Collaborative Retrieval-Augmented Large Language Models Improve Long-tail Recommendationcited by 44
  31. 2023Global and Local Mixture Consistency Cumulative Learning for Long-tailed Visual Recognitionscited by 43
  32. 2024Probabilistic Contrastive Learning for Long-Tailed Visual Recognitioncited by 42
  33. 2021Benign Overfitting in Multiclass Classification: All Roads Lead to Interpolationcited by 41
  34. 2021CoSSL: Co-Learning of Representation and Classifier for Imbalanced Semi-Supervised Learningcited by 41
  35. 2022TAM: Topology-Aware Margin Loss for Class-Imbalanced Node Classificationcited by 41
  36. 2021Graph Few-shot Class-incremental Learningcited by 40
  37. 2021On Model Calibration for Long-Tailed Object Detection and Instance Segmentationcited by 39
  38. 2021Towards Calibrated Model for Long-Tailed Visual Recognition from Prior Perspectivecited by 39

50 most cited of 605 All 605 (JSON)