mlsanityData to 7 Oct 2026

Adversarial Filters of Dataset Biases

arXiv 2002.04108 ↗First posted 10 Feb 2020cs.LG cs.AI cs.CLcited by 159

Not on the map. Why?

Abstract (arXiv)

Large neural models have demonstrated human-level performance on language and vision benchmarks, while their performance degrades considerably on adversarial or out-of-distribution samples. This raises the question of whether these models have learned to solve a dataset rather than the underlying task by overfitting to spurious dataset biases. We investigate one recently proposed approach, AFLite, which adversarially filters such dataset biases, as a means to mitigate the prevalent overestimation of machine performance. We provide a theoretical understanding for AFLite, by situating it in the generalized framework for optimum bias reduction. We present extensive supporting evidence that AFLite is broadly applicable for reduction of measurable dataset biases, and that models trained on the filtered datasets yield better generalization to out-of-distribution tasks. Finally, filtering results in a large drop in model performance (e.g., from 92% to 62% for SNLI), while human performance still remains high. Our work thus shows that such filtered datasets can pose new research challenges for robust generalization by serving as upgraded benchmarks.

Key facts. Cited by 159 papers in the map (data through 7 Oct 2026). First posted on arXiv 10 Feb 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 35; 2022: 47; 2023: 31; 2024: 20; 2025: 15; 2026: 11.

159 papers.

Sort:

By area:

  1. 2023DataComp: In search of the next generation of multimodal datasetscited by 635
  2. 2021Understanding Dataset Difficulty with V-Usable Informationcited by 288
  3. 2021Symbolic Knowledge Distillation: from General Language Models to Commonsense Modelscited by 270
  4. 2023SugarCrepe: Fixing Hackable Benchmarks for Vision-Language Compositionalitycited by 238
  5. 2021Get Your Vitamin C! Robust Fact Verification with Contrastive Evidencecited by 222
  6. 2021What Will it Take to Fix Benchmarking in Natural Language Understanding?cited by 200
  7. 2022WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creationcited by 200
  8. 2021Advances in adversarial attacks and defenses in computer vision: A surveycited by 127
  9. 2022CommonsenseQA 2.0: Exposing the Limits of AI through Gamificationcited by 108
  10. 2021Measure and Improve Robustness in NLP Models: A Surveycited by 107
  11. 2021UNICORN on RAINBOW: A Universal Commonsense Reasoning Model on a New Multitask Benchmarkcited by 98
  12. 2021Challenges in Automated Debiasing for Toxic Language Detectioncited by 95
38 more
  1. 2021Competency Problems: On Finding and Removing Artifacts in Language Datacited by 91
  2. 2021CREAK: A Dataset for Commonsense Reasoning over Entity Knowledgecited by 88
  3. 2021FINE Samples for Learning with Noisy Labelscited by 82
  4. 2021NaturalProofs: Mathematical Theorem Proving in Natural Languagecited by 80
  5. 2022Efficient Methods for Natural Language Processing: A Surveycited by 78
  6. 2021Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Modelscited by 76
  7. 2024SuRe: Summarizing Retrievals using Answer Candidates for Open-domain QA of LLMscited by 67
  8. 2022All You May Need for VQA are Image Captionscited by 65
  9. 2021Mind Your Outliers! Investigating the Negative Impact of Outliers on Active Learning for Visual Question Answeringcited by 62
  10. 2022Handling Bias in Toxic Speech Detection: A Surveycited by 60
  11. 2021TIMEDIAL: Temporal Commonsense Reasoning in Dialogcited by 55
  12. 2022Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasetscited by 54
  13. 2023Explore Spurious Correlations at the Concept Level in Language Models for Text Classificationcited by 48
  14. 2022Don't Blame the Annotator: Bias Already Starts in the Annotation Instructionscited by 47
  15. 2021COM2SENSE: A Commonsense Reasoning Benchmark with Complementary Sentencescited by 40
  16. 2021The Dangers of Underclaiming: Reasons for Caution When Reporting How NLP Systems Failcited by 38
  17. 2022Commonsense Knowledge Reasoning and Generation with Pre-trained Language Models: A Surveycited by 38
  18. 2024Examining the robustness of LLM evaluation to the distributional assumptions of benchmarkscited by 31
  19. 2022DISCO: Distilling Counterfactuals with Large Language Modelscited by 29
  20. 2024Unleashing the Power of Data Tsunami: A Comprehensive Survey on Data Assessment and Selection for Instruction Tuning of Language Modelscited by 29
  21. 2021Finding needles in a haystack: Sampling Structurally-diverse Training Sets from Synthetic Data for Compositional Generalizationcited by 27
  22. 2025Benchmark Designers Should "Train on the Test Set" to Expose Exploitable Non-Visual Shortcutscited by 26
  23. 2023People Make Better Edits: Measuring the Efficacy of LLM-Generated Counterfactually Augmented Data for Harmful Language Detectioncited by 25
  24. 2021How Robust are Model Rankings: A Leaderboard Customization Approach for Equitable Evaluationcited by 24
  25. 2021Analyzing Dynamic Adversarial Training Data in the Limitcited by 24
  26. 2022How Robust is Unsupervised Representation Learning to Distribution Shift?cited by 24
  27. 2022Generalized but not Robust? Comparing the Effects of Data Modification Methods on Out-of-Domain Generalization and Adversarial Robustnesscited by 23
  28. 2022Law Informs Code: A Legal Informatics Approach to Aligning Artificial Intelligence with Humanscited by 23
  29. 2022Mitigating Dataset Bias by Using Per-sample Gradientcited by 21
  30. 2024Towards Trustworthy AI: A Review of Ethical and Robust Large Language Modelscited by 21
  31. 2024The Hard Positive Truth about Vision-Language Compositionalitycited by 21
  32. 2022WinoGAViL: Gamified Association Benchmark to Challenge Vision-and-Language Modelscited by 20
  33. 2023A Survey on Out-of-Distribution Evaluation of Neural NLP Modelscited by 20
  34. 2024Improving Model Evaluation using SMART Filtering of Benchmark Datasetscited by 20
  35. 2021Improving Paraphrase Detection with the Adversarial Paraphrasing Taskcited by 19
  36. 2021Context-aware Adversarial Training for Name Regularity Bias in Named Entity Recognitioncited by 19
  37. 2022On the Limitations of Dataset Balancing: The Lost Battle Against Spurious Correlationscited by 19
  38. 2021DREAM: Improving Situational QA by First Elaborating the Situationcited by 18

50 most cited of 159 All 159 (JSON)