mlsanityData to 13 Oct 2026

On the Accuracy of Influence Functions for Measuring Group Effects

arXiv 1905.13289 ↗First posted 30 May 2019cs.LG stat.MLcited by 146

Not on the map. Why?

Abstract (arXiv)

Influence functions estimate the effect of removing a training point on a model without the need to retrain. They are based on a first-order Taylor approximation that is guaranteed to be accurate for sufficiently small changes to the model, and so are commonly used to study the effect of individual points in large datasets. However, we often want to study the effects of large groups of training points, e.g., to diagnose batch effects or apportion credit between different data sources. Removing such large groups can result in significant changes to the model. Are influence functions still accurate in this setting? In this paper, we find that across many different types of groups and for a range of real-world datasets, the predicted effect (using influence functions) of a group correlates surprisingly well with its actual effect, even if the absolute and relative errors are large. Our theoretical analysis shows that such strong correlation arises only under certain settings and need not hold in general, indicating that real-world datasets have particular properties that allow the influence approximation to be accurate.

Key facts. Cited by 146 papers in the map (data through 13 Oct 2026). First posted on arXiv 30 May 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2021: 18; 2022: 22; 2023: 31; 2024: 25; 2025: 34; 2026: 16.

146 papers.

Sort:

By area:

  1. 2023DataComp: In search of the next generation of multimodal datasetscited by 639
  2. 2023Large Language Model Unlearningcited by 300
  3. 2023TRAK: Attributing Model Behavior at Scalecited by 279
  4. 2021Machine Unlearning of Features and Labelscited by 229
  5. 2021Fast Yet Effective Machine Unlearningcited by 216
  6. 2022Datamodels: Predicting Predictions from Training Datacited by 187
  7. 2021Counterfactual Memorization in Neural Language Modelscited by 155
  8. 2022If Influence Functions are the Answer, Then What is the Question?cited by 143
  9. 2022Training Data Influence Analysis and Estimation: A Surveycited by 134
  10. 2022Dataset Pruning: Reducing Training Data by Examining Generalization Influencecited by 132
  11. 2021Interpretable Deep Learning: Interpretation, Interpretability, Trustworthiness, and Beyondcited by 118
  12. 2023In-context Example Selection with Influencescited by 60
38 more
  1. 2023Simfluence: Modeling the Influence of Individual Training Examples by Simulating Training Runscited by 52
  2. 2022Measuring the Effect of Training Data on Deep Learning Predictions via Randomized Experimentscited by 50
  3. 2021Interactive Label Cleaning with Example-based Explanationscited by 48
  4. 2023Intriguing Properties of Data Attribution on Diffusion Modelscited by 47
  5. 2023The Journey, Not the Destination: How Data Guides Diffusion Modelscited by 42
  6. 2021HYDRA: Hypergradient Data Relevance Analysis for Interpreting Deep Neural Networkscited by 38
  7. 2022Achieving Fairness at No Utility Cost via Data Reweighing with Influencecited by 38
  8. 2022First is Better Than Last for Language Data Influencecited by 34
  9. 2024Advances and Open Challenges in Federated Foundation Modelscited by 34
  10. 2024Influence Functions for Scalable Data Attribution in Diffusion Modelscited by 32
  11. 2024Most Influential Subset Selection: Challenges, Promises, and Beyondcited by 31
  12. 2022Making Recommender Systems Forget: Learning and Unlearning for Erasable Recommendationcited by 29
  13. 2023Theoretical and Practical Perspectives on what Influence Functions Docited by 28
  14. 2022Understanding Rare Spurious Correlations in Neural Networkscited by 27
  15. 2022Repairing Neural Networks by Leaving the Right Past Behindcited by 27
  16. 2021Rethinking Influence Functions of Neural Networks in the Over-parameterized Regimecited by 26
  17. 2022Markov Chain Monte Carlo-Based Machine Unlearning: Unlearning What Needs to be Forgottencited by 25
  18. 2022Understanding Influence Functions and Datamodels via Harmonic Analysiscited by 23
  19. 2025CUPID: Curating Data your Robot Loves with Influence Functionscited by 22
  20. 2021Understanding Instance-based Interpretability of Variational Auto-Encoderscited by 21
  21. 2023Regularizing Second-Order Influences for Continual Learningcited by 21
  22. 2025MAGIC: Near-Optimal Data Attribution for Deep Learningcited by 21
  23. 2023Revisiting the Fragility of Influence Functionscited by 20
  24. 2023Leave-one-out Distinguishability in Machine Learningcited by 20
  25. 2024A CLIP-Powered Framework for Robust and Generalizable Data Selectioncited by 20
  26. 2024Unlearning Traces the Influential Training Data of Language Modelscited by 18
  27. 2021Leveraging Expert Consistency to Improve Algorithmic Decision Supportcited by 17
  28. 2021Adversarial Attacks on Knowledge Graph Embeddings via Instance Attribution Methodscited by 16
  29. 2022Data Debugging with Shapley Importance over End-to-End Machine Learning Pipelinescited by 16
  30. 2022Characterizing the Influence of Graph Elementscited by 16
  31. 2023The Memory Perturbation Equation: Understanding Model's Sensitivity to Datacited by 16
  32. 2023Making Users Indistinguishable: Attribute-wise Unlearning in Recommender Systemscited by 15
  33. 2023Natural Example-Based Explainability: a Surveycited by 13
  34. 2025Group-Level Data Selection for Efficient Pretrainingcited by 13
  35. 2021On Sample Based Explanation Methods for NLP:Efficiency, Faithfulness, and Semantic Evaluationcited by 12
  36. 2022FairIF: Boosting Fairness in Deep Learning via Influence Functions with Validation Set Sensitive Attributescited by 12
  37. 2026GradAlign: Gradient-Aligned Data Selection for LLM Reinforcement Learningcited by 12
  38. 2021Out-of-Distribution Generalization Analysis via Influence Functioncited by 11

50 most cited of 146 All 146 (JSON)