mlsanityData to 14 Oct 2026

SG-OIF: A Stability-Guided Online Influence Framework for Reliable Vision Data

arXiv 2511.19466 ↗First posted 21 Nov 2025cs.CV cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Approximating training-point influence on test predictions is critical for deploying deep-learning vision models, essential for locating noisy data. Though the influence function was proposed for attributing how infinitesimal up-weighting or removal of individual training examples affects model outputs, its implementation is still challenging in deep-learning vision models: inverse-curvature computations are expensive, and training non-stationarity invalidates static approximations. Prior works use iterative solvers and low-rank surrogates to reduce cost, but offline computation lags behind training dynamics, and missing confidence calibration yields fragile rankings that misidentify critical examples. To address these challenges, we introduce a Stability-Guided Online Influence Framework (SG-OIF), the first framework that treats algorithmic stability as a real-time controller, which (i) maintains lightweight anchor IHVPs via stochastic Richardson and preconditioned Neumann; (ii) proposes modular curvature backends to modulate per-example influence scores using stability-guided residual thresholds, anomaly gating, and confidence. Experimental results show that SG-OIF achieves SOTA (State-Of-The-Art) on noise-label and out-of-distribution detection tasks across multiple datasets with various corruption. Notably, our approach achieves 91.1% accuracy in the top 1% prediction samples on the CIFAR-10 (20% asym), and gets 99.8% AUPR score on MNIST, effectively demonstrating that this framework is a practical controller for online influence estimation.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 54 of its 79 references that are in the map. First posted on arXiv 21 Nov 2025.

Builds on

54 of 79 references.

References · 54

  1. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  2. 2017mixup: Beyond Empirical Risk Minimizationcited by 5,183
  3. 2017Understanding Black-box Predictions via Influence Functionscited by 1,741
  4. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,416
  5. 2020Energy-based Out-of-distribution Detectioncited by 1,299
  6. 2019Machine Unlearningcited by 1,061
  7. 2018Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labelscited by 1,059
  8. 2021Towards Total Recall in Industrial Anomaly Detectioncited by 776
  9. 2017MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labelscited by 653
  10. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 618
  11. 2020Personalized Federated Learning with Moreau Envelopescited by 618
  12. 2022Out-of-Distribution Detection with Deep Nearest Neighborscited by 600
38 more
  1. 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 580
  2. 2021CutPaste: Self-Supervised Learning for Anomaly Detection and Localizationcited by 531
  3. 2021DRAEM -- A discriminatively trained reconstruction embedding for surface anomaly detectioncited by 514
  4. 2019Symmetric Cross Entropy for Robust Learning with Noisy Labelscited by 500
  5. 2020What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimationcited by 497
  6. 2021ReAct: Out-of-distribution Detection With Rectified Activationscited by 486
  7. 2019Scaling Out-of-Distribution Detection for Real-World Settingscited by 465
  8. 2020CSI: Novelty Detection via Contrastive Learning on Distributionally Shifted Instancescited by 441
  9. 2019Confident Learning: Estimating Uncertainty in Dataset Labelscited by 413
  10. 2020Generalized ODIN: Detecting Out-of-distribution Image without Learning from Out-of-distribution Datacited by 408
  11. 2022ViM: Out-Of-Distribution with Virtual-logit Matchingcited by 405
  12. 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 366
  13. 2021On the Importance of Gradients for Detecting Distributional Shifts in the Wildcited by 339
  14. 2018Adversarially Robust Generalization Requires More Datacited by 318
  15. 2017Certified Defenses for Data Poisoning Attackscited by 303
  16. 2022VOS: Learning What You Don't Know by Virtual Outlier Synthesiscited by 300
  17. 2017WebVision Database: Visual Learning and Understanding from Web Datacited by 262
  18. 2020Influence Functions in Deep Learning Are Fragilecited by 252
  19. 2022Mitigating Neural Network Overconfidence with Logit Normalizationcited by 252
  20. 2018Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shiftcited by 243
  21. 2021MOS: Towards Scaling Out-of-distribution Detection for Large Semantic Spacecited by 233
  22. 2019Understanding and Utilizing Deep Neural Networks Trained with Noisy Labelscited by 209
  23. 2021DICE: Leveraging Sparsification for Out-of-Distribution Detectioncited by 166
  24. 2021Adversarial Reciprocal Points Learning for Open Set Recognitioncited by 153
  25. 2019On the Accuracy of Influence Functions for Measuring Group Effectscited by 147
  26. 2018Stronger Data Poisoning Attacks Break Data Sanitization Defensescited by 135
  27. 2021OpenGAN: Open-Set Recognition via Open Data Generationcited by 135
  28. 2018Rademacher Complexity for Adversarially Robust Generalizationcited by 132
  29. 2019Adversarial Training Can Hurt Generalizationcited by 132
  30. 2021PixMix: Dreamlike Pictures Comprehensively Improve Safety Measurescited by 131
  31. 2020FastIF: Scalable Influence Functions for Efficient Model Interpretation and Debuggingcited by 120
  32. 2021Semantically Coherent Out-of-Distribution Detectioncited by 107
  33. 2018Understanding the Origins of Bias in Word Embeddingscited by 91
  34. 2019On Second-Order Group Influence Functions for Black-Box Predictionscited by 79
  35. 2023Uncertainty Estimation by Fisher Information-based Evidential Deep Learningcited by 58
  36. 2018Generalization Bounds for Uniformly Stable Algorithmscited by 56
  37. 2020Adversarial Learning Guarantees for Linear Hypotheses and Neural Networkscited by 47
  38. 2020Discriminative Jackknife: Quantifying Uncertainty in Deep Learning via Higher-Order Influence Functionscited by 34

50 most cited of 54 All 54 (JSON)

Built on by

Not cited yet.