mlsanityData to 7 Oct 2026

Will Large-scale Generative Models Corrupt Future Datasets?

arXiv 2211.08095 ↗First posted 15 Nov 2022cs.CVcited by 59

Not on the map. Why?

Abstract (arXiv)

Recently proposed large-scale text-to-image generative models such as DALL·E 2, Midjourney, and StableDiffusion can generate high-quality and realistic images from users' prompts. Not limited to the research community, ordinary Internet users enjoy these generative models, and consequently, a tremendous amount of generated images have been shared on the Internet. Meanwhile, today's success of deep learning in the computer vision field owes a lot to images collected from the Internet. These trends lead us to a research question: "will such generated images impact the quality of future datasets and the performance of computer vision models positively or negatively?" This paper empirically answers this question by simulating contamination. Namely, we generate ImageNet-scale and COCO-scale datasets using a state-of-the-art generative model and evaluate models trained with "contaminated" datasets on various tasks, including image classification and image generation. Throughout experiments, we conclude that generated images negatively affect downstream performance, while the significance depends on tasks and the amount of generated images. The generated datasets and the codes for experiments will be publicly released for future research. Generated datasets and source codes are available from urlhttps://github.com/moskomule/dataset-contamination.

Key facts. Cited by 59 papers in the map (data through 7 Oct 2026). Builds on 51 of its 69 references that are in the map. First posted on arXiv 15 Nov 2022.

Builds on

51 of 69 references.

References · 51

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  4. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  5. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  6. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  7. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  8. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  9. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  10. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  11. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  12. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
38 more
  1. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,539
  2. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  3. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  4. 2022Generative Adversarial Networkscited by 5,603
  5. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,712
  6. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,310
  7. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
  8. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,736
  9. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,595
  10. 2022A ConvNet for the 2020scited by 3,545
  11. 2015Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networkscited by 2,048
  12. 2014CIDEr: Consensus-based Image Description Evaluationcited by 2,027
  13. 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,842
  14. 2015Microsoft COCO Captions: Data Collection and Evaluation Servercited by 1,447
  15. 2019Natural Adversarial Examplescited by 1,401
  16. 2015Flickr30k Entities: Collecting Region-to-Phrase Correspondences for Richer Image-to-Sentence Modelscited by 1,352
  17. 2022Scaling Autoregressive Models for Content-Rich Text-to-Image Generationcited by 1,287
  18. 2021Cascaded Diffusion Models for High Fidelity Image Generationcited by 1,155
  19. 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,012
  20. 2014Deep Visual-Semantic Alignments for Generating Image Descriptionscited by 966
  21. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
  22. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 921
  23. 2016SPICE: Semantic Propositional Image Caption Evaluationcited by 820
  24. 2021Tackling the Generative Learning Trilemma with Denoising Diffusion GANscited by 515
  25. 2022Make-A-Scene: Scene-Based Text-to-Image Generation with Human Priorscited by 487
  26. 2021Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpuscited by 481
  27. 2017Data Augmentation Generative Adversarial Networkscited by 278
  28. 2017WebVision Database: Visual Learning and Understanding from Web Datacited by 262
  29. 2022Easily Accessible Text-to-Image Generation Amplifies Demographic Stereotypes at Large Scalecited by 262
  30. 2018Generative Adversarial Network in Medical Imaging: A Reviewcited by 257
  31. 2019Fourier Spectrum Discrepancies in Deep Network Generated Imagescited by 138
  32. 2019Classification Accuracy Score for Conditional Generative Modelscited by 134
  33. 2020A Survey of Label-noise Representation Learning: Past, Present and Futurecited by 121
  34. 2022The Creativity of Text-to-Image Generationcited by 73
  35. 2021A Closer Look at Fourier Spectrum Discrepancies for CNN-generated Images Detectioncited by 68
  36. 2022LAVIS: A Library for Language-Vision Intelligencecited by 55
  37. 2017A Bayesian Data Augmentation Approach for Learning Deep Modelscited by 50
  38. 2021ERNIE-ViLG: Unified Generative Pre-training for Bidirectional Vision-Language Generationcited by 41

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 9; 2024: 21; 2025: 18; 2026: 11.

59 papers.

Sort:

By area:

  1. 2023Self-Consuming Generative Models Go MADcited by 239
  2. 2024Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Datacited by 143
  3. 2024A Tale of Tails: Model Collapse as a Change of Scaling Lawscited by 127
  4. 2023On the Stability of Iterative Retraining of Generative Models on their own Datacited by 98
  5. 2024Model Collapse Demystified: The Case of Regressioncited by 80
  6. 2024Strong Model Collapsecited by 62
  7. 2023Let's Think Outside the Box: Exploring Leap-of-Thought in Large Language Models with Creative Humor Generationcited by 56
  8. 2024Fairness Feedback Loops: Training on Synthetic Data Amplifies Biascited by 55
  9. 2023Towards Understanding the Interplay of Generative Artificial Intelligence and the Internetcited by 54
  10. 2024Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating Worldcited by 49
  11. 2025The Best Instruction-Tuning Data are Those That Fitcited by 48
  12. 2023LLMDet: A Third Party Large Language Models Generated Text Detection Toolcited by 44
38 more
  1. 2024Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verificationcited by 43
  2. 2023Combining Generative Artificial Intelligence (AI) and the Internet: Heading towards Evolution or Degradation?cited by 41
  3. 2024Self-Consuming Generative Models with Curated Data Provably Optimize Human Preferencescited by 35
  4. 2023EVCap: Retrieval-Augmented Image Captioning with External Visual-Name Memory for Open-World Comprehensioncited by 32
  5. 2025Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challengescited by 30
  6. 2023Human-AI Coevolutioncited by 27
  7. 2023Nepotistically Trained Generative-AI Models Collapsecited by 27
  8. 2024Would Deep Generative Models Amplify Bias in Future Models?cited by 24
  9. 2024Universality of the π2/6 Pathway in Avoiding Model Collapsecited by 24
  10. 2024Rate of Model Collapse in Recursive Trainingcited by 22
  11. 2024Deep Generative Models for 3D Medical Image Synthesiscited by 14
  12. 2024A Survey of Defenses Against AI-Generated Visual Media: Detection,Disruption, and Authenticationcited by 12
  13. 2024Model Collapse in the Self-Consuming Chain of Diffusion Finetuning: A Novel Perspective from Quantitative Trait Modelingcited by 11
  14. 2025Escaping Collapse: The Strength of Weak Data for Large Language Model Trainingcited by 11
  15. 2025When Models Don't Collapse: On the Consistency of Iterative MLEcited by 11
  16. 2025A Causality-aware Paradigm for Evaluating Creativity of Multimodal Large Language Modelscited by 10
  17. 2024Why Does ChatGPT "Delve" So Much? Exploring the Sources of Lexical Overrepresentation in Large Language Modelscited by 8
  18. 2025Why Less is More (Sometimes): A Theory of Data Curationcited by 8
  19. 2025Preventing Model Collapse Under Overparametrization: Optimal Mixing Ratios for Interpolation Learning and Ridge Regressioncited by 7
  20. 2024Synth-Align: Improving Trustworthiness in Vision-Language Model with Synthetic Preference Data Alignmentcited by 6
  21. 2025What happens when generative AI models train recursively on each others' outputs?cited by 5
  22. 2024Dealing with Synthetic Data Contamination in Online Continual Learningcited by 4
  23. 2025Model Misalignment and Language Change: Traces of AI-Associated Language in Unscripted Spoken Englishcited by 4
  24. 2026Learning from Synthetic Data: Limitations of ERMcited by 3
  25. 2024Assessing the Efficacy of Invisible Watermarks in AI-Generated Medical Imagescited by 2
  26. 2024Learning by Surprise: Adaptive Mitigation of Model Collapse in Large Language Modelscited by 2
  27. 2025Limits to AI Growth: The Ecological and Social Consequences of Scalingcited by 2
  28. 2025Recursive Training Loops in LLMs: How training data properties modulate distribution shift in generated data?cited by 2
  29. 2025Emergent Specialization: Rare Token Neurons in Language Modelscited by 2
  30. 2025Convergence and Stability Analysis of Self-Consuming Generative Models with Heterogeneous Human Curationcited by 2
  31. 2026Ambient Dataloops: Generative Models for Dataset Refinementcited by 2
  32. 2026On the Difficulty of Learning a Meta-network for Training Data Selectioncited by 2
  33. 2025SynQuE: Estimating Synthetic Dataset Quality Without Annotationscited by 1
  34. 2026Observations and Remedies for Large Language Model Bias in Self-Consuming Performative Loopcited by 1
  35. 2026Drift and selection in LLM text ecosystemscited by 1
  36. 2024From Obstacles to Resources: Semi-supervised Learning Faces Synthetic Data Contaminationcited by 0
  37. 2024WarCov -- Large multilabel and multimodal dataset from social platformcited by 0
  38. 2025Dataset Creation for Visual Entailment using Generative AIcited by 0

50 most cited of 59 All 59 (JSON)