mlsanityData to 7 Oct 2026

Discovering and Mitigating Visual Biases through Keyword Explanation

arXiv 2301.11104 ↗First posted 26 Jan 2023cs.LG cs.CVcited by 58

Not on the map. Why?

Abstract (arXiv)

Addressing biases in computer vision models is crucial for real-world AI deployments. However, mitigating visual biases is challenging due to their unexplainable nature, often identified indirectly through visualization or sample statistics, which necessitates additional human supervision for interpretation. To tackle this issue, we propose the Bias-to-Text (B2T) framework, which interprets visual biases as keywords. Specifically, we extract common keywords from the captions of mispredicted images to identify potential biases in the model. We then validate these keywords by measuring their similarity to the mispredicted images using a vision-language scoring model. The keyword explanation form of visual bias offers several advantages, such as a clear group naming for bias discovery and a natural extension for debiasing using these group names. Our experiments demonstrate that B2T can identify known biases, such as gender bias in CelebA, background bias in Waterbirds, and distribution shifts in ImageNet-R/C. Additionally, B2T uncovers novel biases in larger datasets, such as Dollar Street and ImageNet. For example, we discovered a contextual bias between "bee" and "flower" in ImageNet. We also highlight various applications of B2T keywords, including debiased training, CLIP prompting, and model comparison.

Key facts. Cited by 58 papers in the map (data through 7 Oct 2026). Builds on 63 of its 92 references that are in the map. First posted on arXiv 26 Jan 2023.

Builds on

63 of 92 references.

References · 63

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  4. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  5. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  6. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
  7. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  8. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,310
  9. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  10. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  11. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  12. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
38 more
  1. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,911
  2. 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,842
  3. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,507
  4. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  5. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  6. 2016Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddingscited by 1,304
  7. 2022CoCa: Contrastive Captioners are Image-Text Foundation Modelscited by 1,265
  8. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  9. 2022Reproducible scaling laws for contrastive language-image learningcited by 1,163
  10. 2017ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseasescited by 969
  11. 2013Domain Generalization via Invariant Feature Representationcited by 622
  12. 2021ClipCap: CLIP Prefix for Image Captioningcited by 557
  13. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  14. 2017CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learningcited by 459
  15. 2021Intriguing Properties of Vision Transformerscited by 446
  16. 2017Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraintscited by 420
  17. 2020Fairness in Machine Learning: A Surveycited by 369
  18. 2020Noise or Signal: The Role of Image Backgrounds in Object Recognitioncited by 342
  19. 2017Functional Map of the Worldcited by 338
  20. 2018Women also Snowboard: Overcoming Bias in Captioning Modelscited by 236
  21. 2020No Subclass Left Behind: Fine-Grained Robustness in Coarse-Grained Classification Problemscited by 208
  22. 2022Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlationscited by 171
  23. 2022Domino: Discovering Systematic Errors with Cross-Modal Embeddingscited by 155
  24. 2019Does Object Recognition Work for Everyone?cited by 149
  25. 2022Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP)cited by 137
  26. 2022Natural Language Descriptions of Deep Visual Featurescited by 136
  27. 2021Evaluating CLIP: Towards Characterization of Broader Capabilities and Downstream Implicationscited by 128
  28. 2021Salient ImageNet: How to discover spurious features in Deep Learning?cited by 114
  29. 2020Estimating Example Difficulty Using Variance of Gradientscited by 96
  30. 2022Distilling Model Failures as Directions in Latent Spacecited by 82
  31. 2022MetaShift: A Dataset of Datasets for Evaluating Contextual Distribution Shifts and Training Conflictscited by 81
  32. 2021The Spotlight: A General Method for Discovering Systematic Errors in Deep Learning Modelscited by 63
  33. 2022Contrastive Adapters for Foundation Model Group Robustnesscited by 62
  34. 2022Data Feedback Loops: Model-driven Amplification of Dataset Biasescited by 59
  35. 2020Understanding Failures of Deep Networks via Robust Feature Extractioncited by 56
  36. 2022Discover and Mitigate Unknown Biases with Debiasing Alternate Networkscited by 47
  37. 2021Discover the Unknown Biased Attribute of an Image Classifiercited by 46
  38. 2022Discovering Bugs in Vision Models using Off-the-shelf Image Generation and Captioningcited by 45

50 most cited of 63 All 63 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 5; 2024: 16; 2025: 25; 2026: 12.

58 papers.

Sort:

By area:

  1. 2024VBench++: Comprehensive and Versatile Benchmark Suite for Video Generative Modelscited by 42
  2. 2023S-CLIP: Semi-supervised Vision-Language Learning using Few Specialist Captionscited by 29
  3. 2024BendVLM: Test-Time Debiasing of Vision-Language Embeddingscited by 17
  4. 2023A Systematic Survey of Prompt Engineering on Vision-Language Foundation Modelscited by 12
  5. 2023PRIME: Prioritizing Interpretability in Failure Mode Extractioncited by 9
  6. 2025Project-Probe-Aggregate: Efficient Fine-Tuning for Group Robustnesscited by 9
  7. 2024A Survey on Responsible Generative AI: What to Generate and What Notcited by 8
  8. 2025On Large Multimodal Models as Open-World Image Classifierscited by 8
  9. 2024LADDER: Language-Driven Slice Discovery and Error Rectification in Vision Classifierscited by 7
  10. 2024Organizing Unstructured Image Collections using Natural Languagecited by 7
  11. 2023Distilling Adversarial Prompts from Safety Benchmarks: Report for the Adversarial Nibbler Challengecited by 6
  12. 2025SpurLens: Automatic Detection of Spurious Cues in Multimodal LLMscited by 6
38 more
  1. 2024Aggregated Attributions for Explanatory Analysis of 3D Segmentation Modelscited by 5
  2. 2024What could go wrong? Discovering and describing failure modes in computer visioncited by 5
  3. 2024BAdd: Bias Mitigation through Bias Additioncited by 5
  4. 2024MAVias: Mitigate any Visual Biascited by 5
  5. 2025ConceptScope: Characterizing Dataset Bias via Disentangled Visual Conceptscited by 5
  6. 2024Say My Name: a Model's Bias Discovery Frameworkcited by 4
  7. 2025Implicit Bias Injection Attacks against Text-to-Image Diffusion Modelscited by 4
  8. 2025Classifier-to-Bias: Toward Unsupervised Automatic Bias Detection for Visual Classifierscited by 4
  9. 2025Debiasing CLIP: Interpreting and Correcting Bias in Attention Headscited by 4
  10. 2025How I Met Your Bias: Investigating Bias Amplification in Diffusion Modelscited by 4
  11. 2025Diffusing DeBias: Synthetic Bias Amplification for Model Debiasingcited by 3
  12. 2025NeurFlow: Interpreting Neural Networks through Neuron Groups and Functional Interactionscited by 3
  13. 2025Group-robust Machine Unlearningcited by 3
  14. 2025Processing and acquisition traces in visual encoders: What does CLIP know about your camera?cited by 3
  15. 2025Open Ad-hoc Categorization with Contextualized Feature Learningcited by 3
  16. 2024Language-guided Detection and Mitigation of Unknown Dataset Biascited by 2
  17. 2025Mitigating Multimodal Hallucinations via Gradient-based Self-Reflectioncited by 2
  18. 2025Automated Detection of Visual Attribute Reliance with a Self-Reflective Agentcited by 2
  19. 2025Representation-Level Counterfactual Calibration for Debiased Zero-Shot Recognitioncited by 2
  20. 2025TRoVe: Discovering Error-Inducing Static Feature Biases in Temporal Vision-Language Modelscited by 2
  21. 2026Stay Fair! Ensuring Group Fairness in Diffusion Models Across Guidance Scalescited by 2
  22. 2023Language-assisted Vision Model Debugger: A Sample-Free Approach to Finding and Fixing Bugscited by 1
  23. 2024Configurable Fairness: Direct Optimization of Parity Metrics via Vision-Language Modelscited by 1
  24. 2024Model Debiasing by Learnable Data Augmentationcited by 1
  25. 2024Curriculum-enhanced GroupDRO: Challenging the Norm of Avoiding Curriculum Learning in Subpopulation Shift Setupscited by 1
  26. 2025Class-Conditional Distribution Balancing for Group Robust Classificationcited by 1
  27. 2025Fine-Grained Class-Conditional Distribution Balancing for Debiased Learningcited by 1
  28. 2025CUBIC: Concept Embeddings for Unsupervised Bias Identification using VLMscited by 1
  29. 2025StarFT: Robust Fine-tuning of Zero-shot Models via Spuriosity Alignmentcited by 1
  30. 2025Hestia: Voxel-Face-Aware Hierarchical Next-Best-View Acquisition for Efficient 3D Reconstructioncited by 1
  31. 2025Suppressing VLM Hallucinations with Spectral Representation Filteringcited by 1
  32. 2025MineTheGap: Automatic Mining of Biases in Text-to-Image Modelscited by 1
  33. 2026BiasEdit: A Training-Free Bias-Detect-and-Edit Framework for Learning Fair Visual Classifierscited by 1
  34. 2024Towards Real-world Debiasing: Rethinking Evaluation, Challenge, and Solutioncited by 0
  35. 2024On the Role of Model Prior in Real-World Inductive Reasoningcited by 0
  36. 2025GH-ESD: Grounded Hypothesis-Driven Error Slice Discovery for Instance-Level Vision Taskscited by 0
  37. 2026A multimodal slice discovery framework for systematic failure detection and explanation in medical image classificationcited by 0
  38. 2026VINO: Video-driven Invariance for Non-contextual Objects via Structural Prior Guided De-contextualizationcited by 0

50 most cited of 58 All 58 (JSON)