mlsanityData to 14 Oct 2026

MoMBS: Mixed-order minibatch sampling enhances model training from diverse-quality images

arXiv 2505.18741 ↗First posted 24 May 2025cs.CV cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Natural images exhibit label diversity (clean vs. noisy) in noisy-labeled image classification and prevalence diversity (abundant vs. sparse) in long-tailed image classification. Similarly, medical images in universal lesion detection (ULD) exhibit substantial variations in image quality, encompassing attributes such as clarity and label correctness. How to effectively leverage training images with diverse qualities becomes a problem in learning deep models. Conventional training mechanisms, such as self-paced curriculum learning (SCL) and online hard example mining (OHEM), relieve this problem by reweighting images with high loss values. Despite their success, these methods still confront two challenges: (i) the loss-based measure of sample hardness is imprecise, preventing optimum handling of different cases, and (ii) there exists under-utilization in SCL or over-utilization OHEM with the identified hard samples. To address these issues, this paper revisits the minibatch sampling (MBS), a technique widely used in deep network training but largely unexplored concerning the handling of diverse-quality training samples. We discover that the samples within a minibatch influence each other during training; thus, we propose a novel Mixed-order Minibatch Sampling (MoMBS) method to optimize the use of training samples with diverse qualities. MoMBS introduces a measure that takes both loss and uncertainty into account to surpass a sole reliance on loss and allows for a more refined categorization of high-loss samples by distinguishing them as either poorly labeled and under represented or well represented and overfitted. We prioritize under represented samples as the main gradient contributors in a minibatch and keep them from the negative influences of poorly labeled or overfitted samples with a mixed-order minibatch sampling design.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 98 of its 165 references that are in the map. First posted on arXiv 24 May 2025.

Builds on

98 of 165 references.

References · 98

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  2. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,448
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,926
  4. 2019Class-Balanced Loss Based on Effective Number of Samplescited by 1,245
  5. 2018Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labelscited by 997
  6. 2019Learning Imbalanced Datasets with Label-Distribution-Aware Margin Losscited by 989
  7. 2020DivideMix: Learning with Noisy Labels as Semi-supervised Learningcited by 692
  8. 2018Learning to Reweight Examples for Robust Deep Learningcited by 675
  9. 2020Long-tail learning via logit adjustmentcited by 618
  10. 2020Learning from Noisy Labels with Deep Neural Networks: A Surveycited by 582
  11. 2016Training Region-based Object Detectors with Online Hard Example Miningcited by 579
  12. 2019Symmetric Cross Entropy for Robust Learning with Noisy Labelscited by 500
38 more
  1. 2017Regularizing Neural Networks by Penalizing Confident Output Distributionscited by 464
  2. 2021Deep Long-Tailed Learning: A Surveycited by 417
  3. 2019BBN: Bilateral-Branch Network with Cumulative Learning for Long-Tailed Visual Recognitioncited by 403
  4. 2020Semi-Supervised Semantic Segmentation with Cross-Consistency Trainingcited by 376
  5. 2019Unsupervised Label Noise Modeling and Loss Correctioncited by 328
  6. 2020Combating noisy labels by agreement: A joint training method with co-regularizationcited by 289
  7. 2020Equalization Loss for Long-Tailed Object Recognitioncited by 285
  8. 2020Normalized Loss Functions for Deep Learning with Noisy Labelscited by 257
  9. 2020Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effectcited by 256
  10. 2021Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotationscited by 256
  11. 2021Improving Calibration for Long-Tailed Recognitioncited by 212
  12. 2019Understanding and Utilizing Deep Neural Networks Trained with Noisy Labelscited by 209
  13. 2020Does label smoothing mitigate label noise?cited by 204
  14. 2021Distribution Alignment: A Unified Framework for Long-tail Visual Recognitioncited by 198
  15. 2020Inf-Net: Automatic COVID-19 Lung Infection Segmentation from CT Imagescited by 184
  16. 2020Disentangling Label Distribution for Long-tailed Visual Recognitioncited by 172
  17. 2015Cost Sensitive Learning of Deep Feature Representations from Imbalanced Datacited by 167
  18. 2019SELF: Learning to Filter Noisy Labels with Self-Ensemblingcited by 164
  19. 2017Learning from Noisy Labels with Distillationcited by 161
  20. 2019NLNL: Negative Learning for Noisy Labelscited by 151
  21. 2019Peer Loss Functions: Learning from Noisy Labels without Knowing Noise Ratescited by 142
  22. 2020Dual T: Reducing Estimation Error for Transition Matrix in Label-noise Learningcited by 142
  23. 2021Contrastive Learning based Hybrid Networks for Long-Tailed Image Classificationcited by 130
  24. 2019Learning From Noisy Labels By Regularized Estimation Of Annotator Confusioncited by 127
  25. 2016Resnet in Resnet: Generalizing Residual Architecturescited by 110
  26. 2019Meta Label Correction for Noisy Label Learningcited by 108
  27. 2020The Devil is in Classification: A Simple Framework for Long-tail Object Detection and Instance Segmentationcited by 108
  28. 2021ResLT: Residual Learning for Long-tailed Recognitioncited by 96
  29. 2015Webly Supervised Learning of Convolutional Networkscited by 95
  30. 2018Masking: A New Perspective of Noisy Supervisioncited by 93
  31. 2019Dynamic Curriculum Learning for Imbalanced Data Classificationcited by 93
  32. 2021Influence-Balanced Loss for Imbalanced Visual Classificationcited by 91
  33. 2020Beyond Class-Conditional Assumption: A Primary Attempt to Combat Instance-Dependent Label Noisecited by 82
  34. 2020A Topological Filter for Learning with Label Noisecited by 76
  35. 2023Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustmentcited by 75
  36. 2020Error-Bounded Correction of Noisy Labelscited by 74
  37. 2018Bayesian Uncertainty Estimation for Batch Normalized Deep Networkscited by 70
  38. 2019Robust Inference via Generative Classifiers for Handling Noisy Labelscited by 68

50 most cited of 98 All 98 (JSON)

Built on by

Not cited yet.