mlsanityData to 7 Oct 2026

Part-Based Models Improve Adversarial Robustness

arXiv 2209.09117 ↗First posted 15 Sep 2022cs.CV cs.CR cs.LGcited by 11

Not on the map. Why?

Abstract (arXiv)

We show that combining human prior knowledge with end-to-end learning can improve the robustness of deep neural networks by introducing a part-based model for object classification. We believe that the richer form of annotation helps guide neural networks to learn more robust features without requiring more samples or larger models. Our model combines a part segmentation model with a tiny classifier and is trained end-to-end to simultaneously segment objects into parts and then classify the segmented object. Empirically, our part-based models achieve both higher accuracy and higher adversarial robustness than a ResNet-50 baseline on all three datasets. For instance, the clean accuracy of our part models is up to 15 percentage points higher than the baseline's, given the same level of robustness. Our experiments indicate that these models also reduce texture bias and yield better robustness against common corruptions and spurious correlations. The code is publicly available at https://github.com/chawins/adv-part-model.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 47 of its 58 references that are in the map. First posted on arXiv 15 Sep 2022.

Builds on

47 of 58 references.

References · 47

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  3. 2020Scaling Laws for Neural Language Modelscited by 6,411
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  5. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 4,993
  6. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,427
  7. 2018Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentationcited by 3,210
  8. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  9. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  10. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,442
  11. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  12. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,259
35 more
  1. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  2. 2018Robustness May Be at Odds with Accuracycited by 873
  3. 2020Fast is better than free: Revisiting adversarial trainingcited by 723
  4. 2019Square Attack: a query-efficient black-box adversarial attack via random searchcited by 682
  5. 2020Overfitting in adversarially robust deep learningcited by 504
  6. 2020Measuring Robustness to Natural Distribution Shifts in Image Classificationcited by 452
  7. 2019On Evaluating Adversarial Robustnesscited by 442
  8. 2019Unlabeled Data Improves Adversarial Robustnesscited by 415
  9. 2019Using Pre-Training Can Improve Model Robustness and Uncertaintycited by 373
  10. 2020Noise or Signal: The Role of Image Backgrounds in Object Recognitioncited by 342
  11. 2018Feature Denoising for Improving Adversarial Robustnesscited by 312
  12. 2017Adversarial Examples for Semantic Segmentation and Object Detectioncited by 275
  13. 2020Uncovering the Limits of Adversarial Training against Norm-Bounded Adversarial Examplescited by 275
  14. 2021Improving Robustness using Generated Datacited by 233
  15. 2014Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Partscited by 230
  16. 2019Improving Adversarial Robustness via Promoting Ensemble Diversitycited by 213
  17. 2021Fixing Data Augmentation to Improve Adversarial Robustnesscited by 201
  18. 2019Adversarial Robustness through Local Linearizationcited by 146
  19. 2020Self-Adaptive Training: beyond Empirical Risk Minimizationcited by 126
  20. 2021PartImageNet: A Large, High-Quality Dataset of Partscited by 107
  21. 2020RayS: A Ray Searching Method for Hard-label Adversarial Attackcited by 94
  22. 2022Generating High Fidelity Data from Low-density Regions using Diffusion Modelscited by 78
  23. 2021On the effectiveness of adversarial training against common corruptionscited by 71
  24. 2015Unsupervised Object Discovery and Localization in the Wild: Part-based Matching with Bottom-up Region Proposalscited by 69
  25. 2019Unsupervised Part-Based Disentangling of Object Shape and Appearancecited by 67
  26. 2018Devil in the Details: Towards Accurate Single and Multiple Human Parsingcited by 64
  27. 2017Joint Multi-Person Pose Estimation and Semantic Part Segmentationcited by 41
  28. 2014Deformable Part Models are Convolutional Neural Networkscited by 35
  29. 2014Articulated Pose Estimation by a Graphical Model with Image Dependent Pairwise Relationscited by 24
  30. 2020Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts datasets for Scene Understandingcited by 14
  31. 2014Actions and Attributes from Wholes and Partscited by 12
  32. 2020Improving the affordability of robustness training for DNNscited by 8
  33. 2020UnMask: Adversarial Detection and Defense Through Robust Feature Alignmentcited by 8
  34. 2018Semantic Part Detection via Matching: Learning to Generalize to Novel Viewpoints from Limited Training Datacited by 4
  35. 2019Rearchitecting Classification Frameworks For Increased Robustnesscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 1; 2023: 3; 2024: 4; 2025: 1; 2026: 2.

11 papers.

  1. 2023PDiscoNet: Semantically consistent part discovery for fine-grained recognitioncited by 21
  2. 2024Subobject-level Image Tokenizationcited by 20
  3. 2022Recognizing Object by Components with Human Prior Knowledge Enhances Adversarial Robustness of Deep Neural Networkscited by 18
  4. 2024PartImageNet++ Dataset: Scaling up Part-based Models for Robust Recognitioncited by 7
  5. 2024ObjectCompose: Evaluating Resilience of Vision-Based Models on Object-to-Background Compositional Changescited by 5
  6. 2025On the Promise for Assurance of Differentiable Neurosymbolic Reasoning Paradigmscited by 3
  7. 2024Edge Detectors Can Make Deep Convolutional Neural Networks More Robustcited by 2
  8. 2026Knowledge-Guided Adversarial Training for Infrared Object Detection via Thermal Radiation Modelingcited by 1
  9. 2023Improving the Robustness of Deep Convolutional Neural Networks Through Feature Learningcited by 0
  10. 2023Preventing Errors in Person Detection: A Part-Based Self-Monitoring Frameworkcited by 0
  11. 2026PartImageNet++ Dataset: Enhancing Visual Models with High-Quality Part Annotationscited by 0