mlsanityData to 14 Oct 2026

Bias in Pruned Vision Models: In-Depth Analysis and Countermeasures

arXiv 2304.12622 ↗First posted 25 Apr 2023cs.CV cs.LGcited by 11

Not on the map. Why?

Abstract (arXiv)

Pruning - that is, setting a significant subset of the parameters of a neural network to zero - is one of the most popular methods of model compression. Yet, several recent works have raised the issue that pruning may induce or exacerbate bias in the output of the compressed model. Despite existing evidence for this phenomenon, the relationship between neural network pruning and induced bias is not well-understood. In this work, we systematically investigate and characterize this phenomenon in Convolutional Neural Networks for computer vision. First, we show that it is in fact possible to obtain highly-sparse models, e.g. with less than 10% remaining weights, which do not decrease in accuracy nor substantially increase in bias when compared to dense models. At the same time, we also find that, at higher sparsities, pruned models exhibit higher uncertainty in their outputs, as well as increased correlations, which we directly link to increased bias. We propose easy-to-use criteria which, based only on the uncompressed model, establish whether bias will increase with pruning, and identify the samples most susceptible to biased predictions post-compression.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 44 of its 57 references that are in the map. First posted on arXiv 25 Apr 2023.

Builds on

44 of 57 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  4. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,845
  5. 2014Deep Learning Face Attributes in the Wildcited by 3,693
  6. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,611
  7. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,262
  8. 2015Learning both Weights and Connections for Efficient Neural Networkscited by 2,031
  9. 2016Equality of Opportunity in Supervised Learningcited by 2,007
  10. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,917
  11. 2021A Survey of Quantization Methods for Efficient Neural Network Inferencecited by 815
  12. 2017To prune, or not to prune: exploring the efficacy of pruning for model compressioncited by 603
32 more
  1. 2017Zero-Shot Learning -- A Comprehensive Evaluation of the Good, the Bad and the Uglycited by 538
  2. 2021Sparsity in Deep Learning: Pruning and growth for efficient inference and training in neural networkscited by 527
  3. 2019Rigging the Lottery: Making All Tickets Winnerscited by 495
  4. 2019The State of Sparsity in Deep Neural Networkscited by 476
  5. 2017Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraintscited by 420
  6. 2020The Lottery Ticket Hypothesis for Pre-trained BERT Networkscited by 275
  7. 2021When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentationscited by 260
  8. 2019Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigationcited by 252
  9. 2020Large image datasets: A pyrrhic win for computer vision?cited by 237
  10. 2021Accelerating Sparse Deep Neural Networkscited by 223
  11. 2019What Do Compressed Deep Neural Networks Forget?cited by 171
  12. 2020Characterising Bias in Compressed Modelscited by 167
  13. 2019Towards Fairer Datasets: Filtering and Balancing the Distribution of the People Subtree in the ImageNet Hierarchycited by 156
  14. 2022Domino: Discovering Systematic Errors with Cross-Modal Embeddingscited by 156
  15. 2020Soft Threshold Weight Reparameterization for Learnable Sparsitycited by 151
  16. 2021Deep Learning Through the Lens of Example Difficultycited by 141
  17. 2022The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Modelscited by 115
  18. 2019Fast Sparse ConvNetscited by 86
  19. 2020The iWildCam 2020 Competition Datasetcited by 86
  20. 2021The Spotlight: A General Method for Discovering Systematic Errors in Deep Learning Modelscited by 63
  21. 2021Lost in Pruning: The Effects of Pruning Neural Networks beyond Test Accuracycited by 62
  22. 2021M-FAC: Efficient Matrix-Free Approximations of Second-Order Informationcited by 54
  23. 2021How Well Do Sparse Imagenet Models Transfer?cited by 50
  24. 2020Exacerbating Algorithmic Bias through Fairness Attackscited by 47
  25. 2021Beyond Preserved Accuracy: Evaluating Loyalty and Robustness of BERT Compressioncited by 38
  26. 2022FairGRAPE: Fairness-aware GRAdient Pruning mEthod for Face Attribute Classificationcited by 30
  27. 2022When does Bias Transfer in Transfer Learning?cited by 23
  28. 2020Going Beyond Classification Accuracy Metrics in Model Compressioncited by 15
  29. 2021Simon Says: Evaluating and Mitigating Bias in Pruned Neural Networks with Knowledge Distillationcited by 15
  30. 2022SEAL : Interactive Tool for Systematic Error Analysis and Labelingcited by 15
  31. 2022GMP*: Well-Tuned Gradual Magnitude Pruning Can Outperform Most BERT-Pruning Methodscited by 15
  32. 2022Recall Distortion in Neural Network Pruning and the Undecayed Pruning Algorithmcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 3; 2024: 4; 2025: 2; 2026: 2.

11 papers.

  1. 2024GPTVQ: The Blessing of Dimensionality for LLM Quantizationcited by 82
  2. 2023SocialCounterfactuals: Probing and Mitigating Intersectional Social Biases in Vision-Language Models with Counterfactual Examplescited by 53
  3. 2024Fairness and Bias Mitigation in Computer Vision: A Surveycited by 13
  4. 2023Accurate Neural Network Pruning Requires Rethinking Sparse Optimizationcited by 10
  5. 2024Bias in Motion: Theoretical Insights into the Dynamics of Bias in SGD Trainingcited by 8
  6. 2025Review of Demographic Fairness in Face Recognitioncited by 7
  7. 2024Model Compression Techniques in Biometrics Applications: A Surveycited by 5
  8. 2026Weight Pruning Amplifies Bias: A Multi-Method Study of Compressed LLMs for Edge AIcited by 2
  9. 2023Integrating Fairness and Model Pruning Through Bi-level Optimizationcited by 1
  10. 2025Downsized and Compromised?: Assessing the Faithfulness of Model Compressioncited by 0
  11. 2026Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Modelscited by 0