mlsanityData to 7 Oct 2026

Exploring the Robustness of In-Context Learning with Noisy Labels

arXiv 2404.18191 ↗First posted 28 Apr 2024cs.CL cs.AI cs.CRcited by 13

Not on the map. Why?

Abstract (arXiv)

Recently, the mysterious In-Context Learning (ICL) ability exhibited by Transformer architectures, especially in large language models (LLMs), has sparked significant research interest. However, the resilience of Transformers' in-context learning capabilities in the presence of noisy samples, prevalent in both training corpora and prompt demonstrations, remains underexplored. In this paper, inspired by prior research that studies ICL ability using simple function classes, we take a closer look at this problem by investigating the robustness of Transformers against noisy labels. Specifically, we first conduct a thorough evaluation and analysis of the robustness of Transformers against noisy labels during in-context learning and show that they exhibit notable resilience against diverse types of noise in demonstration labels. Furthermore, we delve deeper into this problem by exploring whether introducing noise into the training set, akin to a form of data augmentation, enhances such robustness during inference, and find that such noise can indeed improve the robustness of ICL. Overall, our fruitful analysis and findings provide a comprehensive understanding of the resilience of Transformer models against label noises during ICL and provide valuable insights into the research on Transformers in natural language processing. Our code is available at https://github.com/InezYu0928/in-context-learning.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 55 of its 68 references that are in the map. First posted on arXiv 28 Apr 2024.

Builds on

55 of 68 references.

References · 55

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2023GPT-4 Technical Reportcited by 18,132
  3. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  4. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  5. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  6. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  7. 2021On the Opportunities and Risks of Foundation Modelscited by 3,961
  8. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  9. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 2,914
  10. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  11. 2022Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?cited by 1,413
  12. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,259
38 more
  1. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  2. 2021An Explanation of In-context Learning as Implicit Bayesian Inferencecited by 915
  3. 2018Robustness May Be at Odds with Accuracycited by 873
  4. 2022Transformers learn in-context by gradient descentcited by 714
  5. 2022What Can Transformers Learn In-Context? A Case Study of Simple Function Classescited by 700
  6. 2022What learning algorithm is in-context learning? Investigations with linear modelscited by 671
  7. 2016Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approachcited by 663
  8. 2017MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labelscited by 652
  9. 2023Baseline Defenses for Adversarial Attacks Against Aligned Language Modelscited by 608
  10. 2020Learning from Noisy Labels with Deep Neural Networks: A Surveycited by 580
  11. 2017Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methodscited by 542
  12. 2023The Falcon Series of Open Language Modelscited by 526
  13. 2019Symmetric Cross Entropy for Robust Learning with Noisy Labelscited by 499
  14. 2023Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrationscited by 401
  15. 2023Transformers as Statisticians: Provable In-Context Learning with In-Context Algorithm Selectioncited by 291
  16. 2023How Robust is Google's Bard to Adversarial Image Attacks?cited by 177
  17. 2018Learning to Learn from Noisy Labeled Datacited by 160
  18. 2023Fundamental Limitations of Alignment in Large Language Modelscited by 152
  19. 2019NLNL: Negative Learning for Noisy Labelscited by 151
  20. 2023What Makes Good Examples for Visual In-Context Learning?cited by 150
  21. 2018CurriculumNet: Weakly Supervised Learning from Large-Scale Web Imagescited by 126
  22. 2022Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networkscited by 126
  23. 2023The Learnability of In-Context Learningcited by 126
  24. 2023Are Emergent Abilities in Large Language Models just In-Context Learning?cited by 109
  25. 2018Iterative Learning with Open-set Noisy Labelscited by 108
  26. 2019Meta Label Correction for Noisy Label Learningcited by 108
  27. 2023Combating Misinformation in the Age of LLMs: Opportunities and Challengescited by 94
  28. 2023Rethinking Model Ensemble in Transfer-based Adversarial Attackscited by 88
  29. 2020A Unified Approach to Interpreting and Boosting Adversarial Transferabilitycited by 71
  30. 2023Understanding In-Context Learning in Transformers and LLMs by Learning to Learn Discrete Functionscited by 62
  31. 2023Adversarial Demonstration Attacks on Large Language Modelscited by 58
  32. 2023CFA: Class-wise Calibrated Fair Adversarial Trainingcited by 53
  33. 2022Practical Evaluation of Adversarial Robustness via Adaptive Auto Attackcited by 44
  34. 2021Explainable Artificial Intelligence Approaches: A Surveycited by 40
  35. 2023Towards Trustworthy and Aligned Machine Learning: A Data-centric Survey with Causality Perspectivescited by 30
  36. 2020A Survey on Deep Learning with Noisy Labels: How to train your model when you cannot trust on the annotations?cited by 25
  37. 2024On the Duality Between Sharpness-Aware Minimization and Adversarial Trainingcited by 24
  38. 2019An Effective Label Noise Model for DNN Text Classificationcited by 22

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 4; 2025: 5; 2026: 4.

13 papers.

Sort:

By area:

  1. 2024A Theoretical Understanding of Self-Correction through In-context Alignmentcited by 43
  2. 2024On the Noise Robustness of In-Context Learning for Text Generationcited by 15
  3. 2024Characterizing stable regions in the residual stream of LLMscited by 3
  4. 2025Provable Low-Frequency Bias of In-Context Learning of Representationscited by 3
  5. 2024MILE: A Mutation Testing Framework of In-Context Learning Systemscited by 2
  6. 2025Incomplete In-context Learningcited by 2
  7. 2025Retrieval-Augmented Generation as Noisy In-Context Learning: A Unified Theory and Risk Boundscited by 2
  8. 2026In Transformer We Trust? A Perspective on Transformer Architecture Failure Modescited by 1
  9. 2026Transformers Learn Robust In-Context Regression under Distributional Uncertaintycited by 1
  10. 2025Robust In-Context Reinforcement Learning Under Reward Poisoning Attackscited by 0
  11. 2025Not All Pixels Are Equal: Pixel-wise Meta-Learning for Medical Segmentation with Noisy Labelscited by 0
  12. 2026Symmetry Reveals Layerwise Dynamics: How Transformers Perform In-Context Classificationcited by 0
1 more
  1. 2026Model-agnostic information transfer and fusion for classification with label noisecited by 0