mlsanityData to 7 Oct 2026

Puzzle Mix: Exploiting Saliency and Local Statistics for Optimal Mixup

arXiv 2009.06962 ↗First posted 15 Sep 2020cs.LG cs.AI cs.CVcited by 260

Not on the map. Why?

Abstract (arXiv)

While deep neural networks achieve great performance on fitting the training distribution, the learned networks are prone to overfitting and are susceptible to adversarial attacks. In this regard, a number of mixup based augmentation methods have been recently proposed. However, these approaches mainly focus on creating previously unseen virtual examples and can sometimes provide misleading supervisory signal to the network. To this end, we propose Puzzle Mix, a mixup method for explicitly utilizing the saliency information and the underlying statistics of the natural examples. This leads to an interesting optimization problem alternating between the multi-label objective for optimal mixing mask and saliency discounted optimal transport objective. Our experiments show Puzzle Mix achieves the state of the art generalization and the adversarial robustness results compared to other mixup methods on CIFAR-100, Tiny-ImageNet, and ImageNet datasets. The source code is available at https://github.com/snu-mllab/PuzzleMix.

Key facts. Cited by 260 papers in the map (data through 7 Oct 2026). First posted on arXiv 15 Sep 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 41; 2022: 62; 2023: 58; 2024: 58; 2025: 27; 2026: 14.

260 papers.

Sort:

By area:

  1. 2021Rainbow Memory: Continual Learning with a Memory of Diverse Samplescited by 273
  2. 2022Dataset Condensation via Efficient Synthetic-Data Parameterizationcited by 187
  3. 2022LIFT: Language-Interfaced Fine-Tuning for Non-Language Machine Learning Taskscited by 146
  4. 2021Co-Mixup: Saliency Guided Joint Mixup with Supermodular Diversitycited by 127
  5. 2021Mix3D: Out-of-Context Data Augmentation for 3D Scenescited by 126
  6. 2021Interpretable Deep Learning: Interpretation, Interpretability, Trustworthiness, and Beyondcited by 116
  7. 2023Diversify Your Vision Datasets with Automatic Diffusion-Based Augmentationcited by 100
  8. 2022A Comprehensive Survey of Image Augmentation Techniques for Deep Learningcited by 95
  9. 2023MedViT: A Robust Vision Transformer for Generalized Medical Image Classificationcited by 89
  10. 2021TransMix: Attend to Mix for Vision Transformerscited by 79
  11. 2024DiffuseMix: Label-Preserving Data Augmentation with Diffusion Modelscited by 78
  12. 2021SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learningcited by 71
38 more
  1. 2021AutoMix: Unveiling the Power of Mixup for Stronger Classifierscited by 69
  2. 2022CycleMix: A Holistic Strategy for Medical Image Segmentation from Scribble Supervisioncited by 64
  3. 2022MixGen: A New Multi-Modal Data Augmentationcited by 61
  4. 2021Regularization Strategy for Point Cloud via Rigidly Mixed Samplecited by 59
  5. 2021When and How Mixup Improves Calibrationcited by 59
  6. 2022C-Mixup: Improving Generalization in Regressioncited by 57
  7. 2021SmoothMix: Training Confidence-calibrated Smoothed Classifiers for Certified Robustnesscited by 49
  8. 2021Meta-FDMixup: Cross-Domain Few-Shot Learning Guided by Labeled Target Datacited by 48
  9. 2021A Systematic Review of Robustness in Deep Learning for Computer Vision: Mind the gap?cited by 48
  10. 2022TokenMix: Rethinking Image Mixing for Data Augmentation in Vision Transformerscited by 48
  11. 2023Credal Bayesian Deep Learningcited by 45
  12. 2022Rethinking Data Augmentation for Single-source Domain Generalization in Medical Image Segmentationcited by 44
  13. 2021SSMix: Saliency-Based Span Mixup for Text Classificationcited by 39
  14. 2024A Comprehensive Survey on Data Augmentationcited by 38
  15. 2021VideoLT: Large-scale Long-tailed Video Recognitioncited by 35
  16. 2021Object-aware Contrastive Learning for Debiased Scene Representationcited by 34
  17. 2021AlignMixup: Improving Representations By Interpolating Aligned Featurescited by 33
  18. 2023ExcelFormer: A neural network surpassing GBDTs on tabular datacited by 33
  19. 2021MixMo: Mixing Multiple Inputs for Multiple Outputs via Deep Subnetworkscited by 32
  20. 2021Graph Transplant: Node Saliency-Guided Graph Mixup with Local Structure Preservationcited by 30
  21. 2022A Survey of Mix-based Data Augmentation: Taxonomy, Methods, Applications, and Explainabilitycited by 30
  22. 2022Semantic-Aware Representation Blending for Multi-Label Image Recognition with Partial Labelscited by 29
  23. 2022Geodesic Multi-Modal Mixup for Robust Fine-Tuningcited by 29
  24. 2022SelecMix: Debiased Learning by Contradicting-pair Samplingcited by 29
  25. 2021Noisy Feature Mixupcited by 27
  26. 2022A Simple Data Mixing Prior for Improving Self-Supervised Learningcited by 27
  27. 2022A Unified Analysis of Mixed Sample Data Augmentation: A Loss Function Perspectivecited by 27
  28. 2022TokenMixup: Efficient Attention-guided Token-level Data Augmentation for Transformerscited by 26
  29. 2021Survey: Image Mixing and Deleting for Data Augmentationcited by 25
  30. 2024ScribFormer: Transformer Makes CNN Work Better for Scribble-based Medical Image Segmentationcited by 24
  31. 2021It Takes Two to Tango: Mixup for Deep Metric Learningcited by 23
  32. 2023Image Data Augmentation Approaches: A Comprehensive Survey and Future directionscited by 23
  33. 2023GuidedMixup: An Efficient Mixup Strategy Guided by Saliency Mapscited by 23
  34. 2023ScribbleVC: Scribble-supervised Medical Image Segmentation with Vision-Class Embeddingcited by 23
  35. 2021Learning Representational Invariances for Data-Efficient Action Recognitioncited by 22
  36. 2021Boosting Discriminative Visual Representation Learning with Scenario-Agnostic Mixupcited by 22
  37. 2021MUM : Mix Image Tiles and UnMix Feature Tiles for Semi-Supervised Object Detectioncited by 21
  38. 2022Model-Agnostic Augmentation for Accurate Graph Classificationcited by 21

50 most cited of 260 All 260 (JSON)