mlsanityData to 7 Oct 2026

Understanding Instance-Level Label Noise: Disparate Impacts and Treatments

arXiv 2102.05336 ↗First posted 10 Feb 2021cs.LG stat.MLcited by 22

Not on the map. Why?

Abstract (arXiv)

This paper aims to provide understandings for the effect of an over-parameterized model, e.g. a deep neural network, memorizing instance-dependent noisy labels. We first quantify the harms caused by memorizing noisy instances, and show the disparate impacts of noisy labels for sample instances with different representation frequencies. We then analyze how several popular solutions for learning with noisy labels mitigate this harm at the instance level. Our analysis reveals that existing approaches lead to disparate treatments when handling noisy instances. While higher-frequency instances often enjoy a high probability of an improvement by applying these solutions, lower-frequency instances do not. Our analysis reveals new understandings for when these approaches work, and provides theoretical justifications for previously reported empirical observations. This observation requires us to rethink the distribution of label noise across instances and calls for different treatments for instances in different regimes.

Key facts. Cited by 22 papers in the map (data through 7 Oct 2026). Builds on 27 of its 40 references that are in the map. First posted on arXiv 10 Feb 2021.

Builds on

27 of 40 references.

References · 27

  1. 2016Understanding deep learning requires rethinking generalizationcited by 1,657
  2. 2018Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labelscited by 1,053
  3. 2018Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labelscited by 994
  4. 2019When Does Label Smoothing Help?cited by 985
  5. 2017A Closer Look at Memorization in Deep Networkscited by 936
  6. 2016Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approachcited by 663
  7. 2017MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labelscited by 652
  8. 2017Exploring Generalization in Deep Learningcited by 595
  9. 2019Symmetric Cross Entropy for Robust Learning with Noisy Labelscited by 499
  10. 2020What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimationcited by 492
  11. 2019Does Learning Require Memorization? A Short Tale about a Long Tailcited by 452
  12. 2019How does Disagreement Help Generalization against Label Corruption?cited by 390
15 more
  1. 2020Combating noisy labels by agreement: A joint training method with co-regularizationcited by 289
  2. 2020Normalized Loss Functions for Deep Learning with Noisy Labelscited by 257
  3. 2019Understanding and Utilizing Deep Neural Networks Trained with Noisy Labelscited by 209
  4. 2019Are Anchor Points Really Indispensable in Label-Noise Learning?cited by 208
  5. 2020Does label smoothing mitigate label noise?cited by 204
  6. 2020Learning with Instance-Dependent Label Noise: A Sample Sieve Approachcited by 158
  7. 2019Peer Loss Functions: Learning from Noisy Labels without Knowing Noise Ratescited by 142
  8. 2020Dual T: Reducing Estimation Error for Transition Matrix in Label-noise Learningcited by 142
  9. 2021Provably End-to-end Label-Noise Learning without Anchor Pointscited by 97
  10. 2020A Second-Order Approach to Learning with Instance-Dependent Label Noisecited by 90
  11. 2021Learning Noise Transition Matrix from Only Noisy Labels via Total Variation Regularizationcited by 62
  12. 2017Learning with Bounded Instance- and Label-dependent Label Noisecited by 57
  13. 2019On Symmetric Losses for Learning from Corrupted Labelscited by 53
  14. 2019How does Early Stopping Help Generalization against Label Noise?cited by 41
  15. 2020Part-dependent Label Noise: Towards Instance-dependent Label Noisecited by 40

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 5; 2022: 5; 2023: 5; 2024: 3; 2025: 3; 2026: 1.

22 papers.

Sort:

  1. 2021To Smooth or Not? When Label Smoothing Meets Noisy Labelscited by 70
  2. 2021Detecting Corrupted Labels Without Training a Model to Predictcited by 57
  3. 2022Identifiability of Label Noise Transition Matrixcited by 34
  4. 2022To Aggregate or Not? Learning with Separate Noisy Labelscited by 29
  5. 2021The Rich Get Richer: Disparate Impact of Semi-Supervised Learningcited by 28
  6. 2022Beyond Images: Label Noise Transition Matrix Estimation for Tasks with Lower-Quality Featurescited by 27
  7. 2021Mitigating Memorization of Noisy Labels via Regularization between Representationscited by 26
  8. 2022Instance-Dependent Noisy Label Learning via Graphical Modellingcited by 17
  9. 2022Weak Proxies are Sufficient and Preferable for Fairness with Missing Sensitive Attributescited by 17
  10. 2024RobustTSF: Towards Theory and Design of Robust Time Series Forecasting with Anomaliescited by 17
  11. 2021Can Less be More? When Increasing-to-Balancing Label Noise Rates Considered Beneficialcited by 13
  12. 2023The Importance of Human-Labeled Data in the Era of LLMscited by 8
10 more
  1. 2023Fairness Improves Learning from Noisily Labeled Long-Tailed Datacited by 5
  2. 2023Leveraging an Alignment Set in Tackling Instance-Dependent Label Noisecited by 4
  3. 2024Learning under Temporal Label Noisecited by 4
  4. 2024For Better or For Worse? Learning Minimum Variance Features With Label Augmentationcited by 3
  5. 2025Regretful Decisions under Label Noisecited by 1
  6. 2023On the Cause of Unfairness: A Training Sample Perspectivecited by 0
  7. 2023FPR Estimation for Fraud Detection in the Presence of Class-Conditional Label Noisecited by 0
  8. 2025Variational Learning Induces Adaptive Label Smoothingcited by 0
  9. 2025Label Smoothing is a Pragmatic Information Bottleneckcited by 0
  10. 2026Analyzing Training-Free Corruption Detection for Object Detection Datasetscited by 0