mlsanityData to 14 Oct 2026

Are All Training Examples Created Equal? An Empirical Study

arXiv 1811.12569 ↗First posted 30 Nov 2018cs.LG cs.CV stat.MLcited by 16

Not on the map. Why?

Abstract (arXiv)

Modern computer vision algorithms often rely on very large training datasets. However, it is conceivable that a carefully selected subsample of the dataset is sufficient for training. In this paper, we propose a gradient-based importance measure that we use to empirically analyze relative importance of training images in four datasets of varying complexity. We find that in some cases, a small subsample is indeed sufficient for training. For other datasets, however, the relative differences in importance are negligible. These results have important implications for active learning on deep networks. Additionally, our analysis method can be used as a general tool to better understand diversity of training examples in datasets.

Key facts. Cited by 16 papers in the map (data through 14 Oct 2026). First posted on arXiv 30 Nov 2018.

Builds on

No reference list yet.

Built on by

Cited by year

Citing papers per year: 2021: 5; 2022: 3; 2023: 4; 2024: 3; 2025: 1.

16 papers.

Sort:

  1. 2021Understanding Dataset Difficulty with V-Usable Informationcited by 291
  2. 2021Representation Matters: Assessing the Importance of Subgroup Allocations in Training Datacited by 31
  3. 2023Towards a statistical theory of data selection under weak supervisioncited by 27
  4. 2021Algorithm Fairness in AI for Medicine and Healthcarecited by 21
  5. 2024Sketchy Moment Matching: Toward Fast and Provable Data Selection for Finetuningcited by 5
  6. 2021One Backward from Ten Forward, Subsampling for Large-Scale Deep Learningcited by 4
  7. 2022Improving Data Quality with Training Dynamics of Gradient Boosting Decision Treescited by 3
  8. 2022Reduce, Reuse, Recycle: Improving Training Efficiency with Distillationcited by 3
  9. 2023Exploring Data Redundancy in Real-world Image Classification through Data Selectioncited by 3
  10. 2021Efficient Visual Recognition with Deep Neural Networks: A Survey on Recent Advances and New Directionscited by 2
  11. 2022How Low Can We Go? Pixel Annotation for Semantic Segmentationcited by 0
  12. 2023Semantically Redundant Training Data Removal and Deep Model Classification Performance: A Study with Chest X-rayscited by 0
4 more
  1. 2023On the Importance of Large Objects in CNN Based Object Detection Algorithmscited by 0
  2. 2024Automatic Pruning of Fine-tuning Datasets for Transformer-based Language Modelscited by 0
  3. 2024textitGeoHard: Towards Measuring Class-wise Hardness through Modelling Class Semanticscited by 0
  4. 2025AGI-Elo: How Far Are We From Mastering A Task?cited by 0