mlsanityData to 7 Oct 2026

Unsupervised Label Noise Modeling and Loss Correction

arXiv 1904.11238 ↗First posted 25 Apr 2019cs.CVcited by 327

Not on the map. Why?

Abstract (arXiv)

Despite being robust to small amounts of label noise, convolutional neural networks trained with stochastic gradient methods have been shown to easily fit random labels. When there are a mixture of correct and mislabelled targets, networks tend to fit the former before the latter. This suggests using a suitable two-component mixture model as an unsupervised generative model of sample loss values during training to allow online estimation of the probability that a sample is mislabelled. Specifically, we propose a beta mixture to estimate this probability and correct the loss by relying on the network prediction (the so-called bootstrapping loss). We further adapt mixup augmentation to drive our approach a step further. Experiments on CIFAR-10/100 and TinyImageNet demonstrate a robustness to label noise that substantially outperforms recent state-of-the-art. Source code is available at https://git.io/fjsvE

Key facts. Cited by 327 papers in the map (data through 7 Oct 2026). First posted on arXiv 25 Apr 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 80; 2022: 82; 2023: 67; 2024: 45; 2025: 27; 2026: 26.

327 papers.

Sort:

By area:

  1. 2023Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervisioncited by 445
  2. 2021Understanding Dataset Difficulty with V-Usable Informationcited by 288
  3. 2021Understanding and Improving Early Stopping for Learning with Noisy Labelscited by 112
  4. 2022Selective-Supervised Contrastive Learning with Noisy Labelscited by 112
  5. 2021Learning with Feature-Dependent Label Noise: A Progressive Approachcited by 102
  6. 2021Augmentation Strategies for Learning with Noisy Labelscited by 86
  7. 2021Open-Vocabulary Instance Segmentation via Robust Cross-Modal Pseudo-Labelingcited by 84
  8. 2023Provable Dynamic Fusion for Low-Quality Multimodal Datacited by 81
  9. 2021Jo-SRC: A Contrastive Approach for Combating Noisy Labelscited by 80
  10. 2021Robust Audio-Visual Instance Discriminationcited by 80
  11. 2021Sample Selection with Uncertainty of Losses for Learning with Noisy Labelscited by 79
  12. 2022Source-Free Unsupervised Domain Adaptation: A Surveycited by 79
38 more
  1. 2022UNICON: Combating Label Noise Through Uniform Selection and Contrastive Learningcited by 77
  2. 2021Contrast to Divide: Self-Supervised Pre-Training for Learning with Noisy Labelscited by 69
  3. 2021Learning Fast Sample Re-weighting Without Reward Datacited by 69
  4. 2022Learn From All: Erasing Attention Consistency for Noisy Label Facial Expression Recognitioncited by 67
  5. 2022FedCorr: Multi-Stage Federated Learning for Label Noise Correctioncited by 61
  6. 2021Open-set Label Noise Can Improve Robustness Against Inherent Label Noisecited by 58
  7. 2021NGC: A Unified Framework for Learning with Open-World Noisy Datacited by 58
  8. 2023Gradient Norm Aware Minimization Seeks First-Order Flatness and Improves Generalizationcited by 57
  9. 2022Learning with Neighbor Consistency for Noisy Labelscited by 55
  10. 2021Co-learning: Learning from Noisy Labels with Self-supervisioncited by 50
  11. 2021Sample Selection for Fair and Robust Trainingcited by 48
  12. 2022ProMix: Combating Label Noise via Maximizing Clean Sample Utilitycited by 48
  13. 2021MetaCorrection: Domain-aware Meta Loss Correction for Unsupervised Domain Adaptation in Semantic Segmentationcited by 43
  14. 2022Eliciting and Learning with Soft Labels from Every Annotatorcited by 43
  15. 2024VariErr NLI: Separating Annotation Error from Human Label Variationcited by 42
  16. 2021LongReMix: Robust Learning with High Confidence Samples in a Noisy Label Environmentcited by 40
  17. 2022Mitigating Memorization of Noisy Labels by Clipping the Model Predictioncited by 40
  18. 2023Towards Open-Set Test-Time Adaptation Utilizing the Wisdom of Crowds in Entropy Minimizationcited by 39
  19. 2021Faster Meta Update Strategy for Noise-Robust Deep Learningcited by 38
  20. 2022Towards Robust Ranker for Text Retrievalcited by 37
  21. 2021Contrastive Learning Improves Model Robustness Under Label Noisecited by 36
  22. 2021Towards Understanding Deep Learning from Noisy Labels with Small-Loss Criterioncited by 36
  23. 2021Joint Negative and Positive Learning for Noisy Labelscited by 34
  24. 2021Gradient Imitation Reinforcement Learning for Low Resource Relation Extractioncited by 33
  25. 2022Self-Filtering: A Noise-Aware Sample Selection for Label Noise with Confidence Penalizationcited by 33
  26. 2022Online Continual Learning on a Contaminated Data Stream with Blurry Task Boundariescited by 31
  27. 2021Simple and Robust Loss Design for Multi-Label Learning with Missing Labelscited by 30
  28. 2022Not All Models Are Equal: Predicting Model Transferability in a Self-challenging Fisher Spacecited by 30
  29. 2022Self-Distillation from the Last Mini-Batch for Consistency Regularizationcited by 29
  30. 2022Joint-Modal Label Denoising for Weakly-Supervised Audio-Visual Video Parsingcited by 29
  31. 2021Noisy-Labeled NER with Confidence Estimationcited by 28
  32. 2022Uncertainty-Aware Learning Against Label Noise on Imbalanced Datasetscited by 28
  33. 2023Twin Contrastive Learning with Noisy Labelscited by 28
  34. 2021Continual Learning on Noisy Data Streams via Self-Purified Replaycited by 27
  35. 2023BiCro: Noisy Correspondence Rectification for Multi-modality Data via Bi-directional Cross-modal Similarity Consistencycited by 27
  36. 2022Metadata Archaeology: Unearthing Data Subsets by Leveraging Training Dynamicscited by 26
  37. 2023Noisy Correspondence Learning with Meta Similarity Correctioncited by 26
  38. 2023From Shortcuts to Triggers: Backdoor Defense with Denoised PoEcited by 26

50 most cited of 327 All 327 (JSON)