mlsanityData to 14 Oct 2026

Unified Risk Analysis for Weakly Supervised Learning

arXiv 2309.08216 ↗First posted 15 Sep 2023cs.LGcited by 5

Not on the map. Why?

Abstract (arXiv)

Among the flourishing research of weakly supervised learning (WSL), we recognize the lack of a unified interpretation of the mechanism behind the weakly supervised scenarios, let alone a systematic treatment of the risk rewrite problem, a crucial step in the empirical risk minimization approach. In this paper, we introduce a framework providing a comprehensive understanding and a unified methodology for WSL. The formulation component of the framework, leveraging a contamination perspective, provides a unified interpretation of how weak supervision is formed and subsumes fifteen existing WSL settings. The induced reduction graphs offer comprehensive connections over WSLs. The analysis component of the framework, viewed as a decontamination process, provides a systematic method of conducting risk rewrite. In addition to the conventional inverse matrix approach, we devise a novel strategy called marginal chain aiming to decontaminate distributions. We justify the feasibility of the proposed framework by recovering existing rewrites reported in the literature.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Builds on 40 of its 57 references that are in the map. First posted on arXiv 15 Sep 2023.

Builds on

40 of 57 references.

References · 40

  1. 2018Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labelscited by 1,059
  2. 2016Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approachcited by 665
  3. 2017MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labelscited by 653
  4. 2019Confident Learning: Estimating Uncertainty in Dataset Labelscited by 413
  5. 2019How does Disagreement Help Generalization against Label Corruption?cited by 390
  6. 2014Training Deep Neural Networks on Noisy Labels with Bootstrappingcited by 322
  7. 2018Joint Optimization Framework for Learning with Noisy Labelscited by 293
  8. 2017Positive-Unlabeled Learning with Non-Negative Risk Estimatorcited by 219
  9. 2018Dimensionality-Driven Learning with Noisy Labelscited by 173
  10. 2020Progressive Identification of True Labels for Partial-Label Learningcited by 102
  11. 2017Learning with Biased Complementary Labelscited by 93
  12. 2020Provably Consistent Partial-Label Learningcited by 92
28 more
  1. 2017Learning from Complementary Labelscited by 88
  2. 2013Classification with Asymmetric Label Noise: Consistency and Maximal Denoisingcited by 68
  3. 2020Confidence Scores Make Instance-dependent Label-noise Learning Possiblecited by 63
  4. 2018Complementary-Label Learning for Arbitrary Losses and Modelscited by 57
  5. 2020Classification with Rejection Based on Cost-sensitive Classificationcited by 55
  6. 2019On Symmetric Losses for Learning from Corrupted Labelscited by 53
  7. 2019Learning with Multiple Complementary Labelscited by 53
  8. 2017Learning with Confident Examples: Rank Pruning for Robust Classification with Noisy Labelscited by 52
  9. 2014Training Convolutional Networks with Noisy Labelscited by 48
  10. 2018On the Minimal Supervision for Training Any Binary Classifier from Only Unlabeled Datacited by 43
  11. 2016Theoretical Comparisons of Positive-Unlabeled Learning against Positive-Negative Learningcited by 41
  12. 2018Classification from Pairwise Similarity and Unlabeled Datacited by 36
  13. 2019Mitigating Overfitting in Supervised Classification from Two Unlabeled Datasets: A Consistent Risk Correction Approachcited by 29
  14. 2020Unbiased Risk Estimators Can Mislead: A Case Study of Learning with Complementary Labelscited by 29
  15. 2022Is the Performance of My Deep Network Too Good to Be True? A Direct Approach to Estimating the Bayes Error in Binary Classificationcited by 26
  16. 2019Classification from Pairwise Similarities/Dissimilarities and Unlabeled Data via Empirical Risk Minimizationcited by 21
  17. 2017Semi-Supervised AUC Optimization based on Positive-Unlabeled Learningcited by 20
  18. 2020Pointwise Binary Classification with Pairwise Confidence Comparisonscited by 20
  19. 2021Learning from Similarity-Confidence Datacited by 19
  20. 2021Binary Classification from Multiple Unlabeled Datasets via Surrogate Set Classificationcited by 12
  21. 2020Bridging Ordinary-Label Learning and Complementary-Label Learningcited by 11
  22. 2016Efficient Training for Positive Unlabeled Learningcited by 10
  23. 2017Decontamination of Mutual Contamination Modelscited by 8
  24. 2021Learning with Proper Partial Labelscited by 8
  25. 2013Clustering Unclustered Data: Unsupervised Binary Labeling of Two Datasets Having Different Class Balancescited by 4
  26. 2019Learning from Multiple Corrupted Sources, with Application to Learning from Label Proportionscited by 4
  27. 2015Learning in the Presence of Corruptioncited by 3
  28. 2021Multi-Class Classification from Single-Class Data with Confidencescited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 2; 2025: 1; 2026: 2.

5 papers.

  1. 2024Theoretical Analysis of Weak-to-Strong Generalizationcited by 48
  2. 2024A General Framework for Learning from Weak Supervisioncited by 7
  3. 2025LLM Routing with Dueling Feedbackcited by 7
  4. 2026Risk-Consistent Multiclass Learning from Random Label-Subset Membership Queriescited by 0
  5. 2026Importance Weighting for Unlabeled-unlabeled Learning under Distribution Shiftcited by 0