mlsanityData to 7 Oct 2026

Affinity and Diversity: Quantifying Mechanisms of Data Augmentation

arXiv 2002.08973 ↗First posted 20 Feb 2020cs.LG cs.CV stat.MLcited by 45

Not on the map. Why?

Abstract (arXiv)

Though data augmentation has become a standard component of deep neural network training, the underlying mechanism behind the effectiveness of these techniques remains poorly understood. In practice, augmentation policies are often chosen using heuristics of either distribution shift or augmentation diversity. Inspired by these, we seek to quantify how data augmentation improves model generalization. To this end, we introduce interpretable and easy-to-compute measures: Affinity and Diversity. We find that augmentation performance is predicted not by either of these alone but by jointly optimizing the two.

Key facts. Cited by 45 papers in the map (data through 7 Oct 2026). First posted on arXiv 20 Feb 2020.

Builds on

No reference list yet.

Built on by

Cited by year

Citing papers per year: 2021: 13; 2022: 9; 2023: 8; 2024: 9; 2025: 6.

45 papers.

Sort:

By area:

  1. 2021A Survey of Data Augmentation Approaches for NLPcited by 408
  2. 2021SCARF: Self-Supervised Contrastive Learning using Random Feature Corruptioncited by 151
  3. 2022How Much Data Are Augmentations Worth? An Investigation into Scaling Laws, Invariance, and Implicit Regularizationcited by 40
  4. 2024A Comprehensive Survey on Data Augmentationcited by 38
  5. 2021MixMo: Mixing Multiple Inputs for Multiple Outputs via Deep Subnetworkscited by 32
  6. 2021Augmentations in Graph Contrastive Learning: Current Methodological Flaws & Towards Better Practicescited by 24
  7. 2022The good, the bad and the ugly sides of data augmentation: An implicit spectral regularization perspectivecited by 24
  8. 2023Out-of-Domain Robustness via Targeted Augmentationscited by 24
  9. 2022A Theory of PAC Learnability under Transformation Invariancescited by 21
  10. 2021Barely-Supervised Learning: Semi-Supervised Learning with very few labeled imagescited by 17
  11. 2021DivAug: Plug-in Automated Data Augmentation with Explicit Diversity Maximizationcited by 15
  12. 2021CutDepth:Edge-aware Data Augmentation in Depth Estimationcited by 14
33 more
  1. 2023IPMix: Label-Preserving Data Augmentation Method for Training Robust Classifierscited by 14
  2. 2022A Survey of Automated Data Augmentation Algorithms for Deep Learning-based Image Classification Taskscited by 12
  3. 2023Mask-then-Fill: A Flexible and Effective Data Augmentation Framework for Event Extractioncited by 12
  4. 2023Understanding the Detrimental Class-level Effects of Data Augmentationcited by 12
  5. 2024Chasing Progress, Not Perfection: Revisiting Strategies for End-to-End LLM Plan Generationcited by 11
  6. 2021Local Patch AutoAugment with Multi-Agent Collaborationcited by 8
  7. 2021The Imaginative Generative Adversarial Network: Automatic Data Augmentation for Dynamic Skeleton-Based Hand Gesture and Human Action Recognitioncited by 8
  8. 2024A Survey of Data Synthesis Approachescited by 8
  9. 2023Boosting Model Inversion Attacks with Adversarial Examplescited by 7
  10. 2025Diversity-oriented Data Augmentation with Large Language Modelscited by 7
  11. 2022How to Fine-tune Models with Few Samples: Update, Data Augmentation, and Test-time Augmentationcited by 6
  12. 2023Semi-Supervised Learning for Multi-Label Cardiovascular Diseases Prediction:A Multi-Dataset Studycited by 6
  13. 2021Principled Ultrasound Data Augmentation for Classification of Standard Planescited by 5
  14. 2023Domain Generalization by Rejecting Extreme Augmentationscited by 5
  15. 2024Local Superior Soups: A Catalyst for Model Merging in Cross-Silo Federated Learningcited by 5
  16. 2025Measuring Diversity in Synthetic Datasetscited by 5
  17. 2021On Data-centric Mythscited by 4
  18. 2024Data Diversity as Implicit Regularization: How Does Diversity Shape the Weight Space of Deep Neural Networks?cited by 4
  19. 2024A Theoretical Characterization of Optimal Data Augmentations in Self-Supervised Learningcited by 4
  20. 2025Data Can Speak for Itself: Quality-guided Utilization of Wireless Synthetic Datacited by 4
  21. 2022VITA: A Multi-Source Vicinal Transfer Augmentation Method for Out-of-Distribution Generalizationcited by 3
  22. 2022ReSmooth: Detecting and Utilizing OOD Samples when Training with Data Augmentationcited by 3
  23. 2022Semantically Consistent Data Augmentation for Neural Machine Translation via Conditional Masked Language Modelcited by 3
  24. 2021On the Effects of Artificial Data Modificationcited by 2
  25. 2023MixEdit: Revisiting Data Augmentation and Beyond for Grammatical Error Correctioncited by 2
  26. 2024Relative Counterfactual Contrastive Learning for Mitigating Pretrained Stance Bias in Stance Detectioncited by 2
  27. 2024Transesophageal Echocardiography Generation using Anatomical Modelscited by 2
  28. 2021CGCL: Collaborative Graph Contrastive Learning without Handcrafted Graph Data Augmentationscited by 1
  29. 2025Data-Efficient Self-Supervised Algorithms for Fine-Grained Birdsong Analysiscited by 1
  30. 2022Generalized Supervised Contrastive Learningcited by 0
  31. 2024Trimming the Risk: Towards Reliable Continuous Training for Deep Learning Inspection Systemscited by 0
  32. 2025LayerMix: Enhanced Data Augmentation through Fractal Integration for Robust Deep Learningcited by 0
  33. 2025Revisiting Mixout: An Overlooked Path to Robust Finetuningcited by 0