mlsanityData to 7 Oct 2026

Pixel-Grounded Prototypical Part Networks

arXiv 2309.14531 ↗First posted 25 Sep 2023cs.CVcited by 18

Not on the map. Why?

Abstract (arXiv)

Prototypical part neural networks (ProtoPartNNs), namely PROTOPNET and its derivatives, are an intrinsically interpretable approach to machine learning. Their prototype learning scheme enables intuitive explanations of the form, this (prototype) looks like that (testing image patch). But, does this actually look like that? In this work, we delve into why object part localization and associated heat maps in past work are misleading. Rather than localizing to object parts, existing ProtoPartNNs localize to the entire image, contrary to generated explanatory visualizations. We argue that detraction from these underlying issues is due to the alluring nature of visualizations and an over-reliance on intuition. To alleviate these issues, we devise new receptive field-based architectural constraints for meaningful localization and a principled pixel space mapping for ProtoPartNNs. To improve interpretability, we propose additional architectural improvements, including a simplified classification head. We also make additional corrections to PROTOPNET and its derivatives, such as the use of a validation set, rather than a test set, to evaluate generalization during training. Our approach, PIXPNET (Pixel-grounded Prototypical part Network), is the only ProtoPartNN that truly learns and localizes to prototypical object parts. We demonstrate that PIXPNET achieves quantifiably improved interpretability without sacrificing accuracy.

Key facts. Cited by 18 papers in the map (data through 7 Oct 2026). Builds on 52 of its 95 references that are in the map. First posted on arXiv 25 Sep 2023.

Builds on

52 of 95 references.

References · 52

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  4. 2016Densely Connected Convolutional Networkscited by 6,691
  5. 2013Intriguing properties of neural networkscited by 5,114
  6. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 4,993
  7. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,507
  8. 2019Natural Adversarial Examplescited by 1,401
  9. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,400
  10. 2018This Looks Like That: Deep Learning for Interpretable Image Recognitioncited by 731
  11. 2016The Mythos of Model Interpretabilitycited by 716
  12. 2018Do Better ImageNet Models Transfer Better?cited by 560
38 more
  1. 2017Understanding the Effective Receptive Field in Deep Convolutional Neural Networkscited by 549
  2. 2019Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methodscited by 440
  3. 2020NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Searchcited by 412
  4. 2019A Survey on Explainable Artificial Intelligence (XAI): Towards Medical XAIcited by 403
  5. 2019Explanations can be manipulated and geometry is to blamecited by 215
  6. 2020Neural Prototype Trees for Interpretable Fine-grained Image Recognitioncited by 213
  7. 2018Do CIFAR-10 Classifiers Generalize to CIFAR-10?cited by 197
  8. 2022The Disagreement Problem in Explainable Machine Learning: A Practitioner's Perspectivecited by 156
  9. 2021HIVE: Evaluating the Human Interpretability of Visual Explanationscited by 105
  10. 2021Interpretable Image Classification with Differentiable Prototypes Assignmentcited by 97
  11. 2020Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Databasecited by 95
  12. 2020On the Tractability of SHAP Explanationscited by 92
  13. 2021A Comprehensive Taxonomy for Explainable Artificial Intelligence: A Systematic Survey of Surveys on Methods and Conceptscited by 90
  14. 2020Explaining the Explainer: A First Theoretical Analysis of LIMEcited by 89
  15. 2020ProtoPShare: Prototype Sharing for Interpretable Image Classification and Similarity Discoverycited by 89
  16. 2019Interpretable and Steerable Sequence Learning via Prototypescited by 72
  17. 2021XProtoNet: Diagnosis in Chest Radiography with Global and Local Explanationscited by 72
  18. 2021This Looks Like That... Does it? Shortcomings of Latent Space Prototype Interpretability in Deep Networkscited by 65
  19. 2022ProtoPFormer: Concentrating on Prototypical Parts in Vision Transformers for Interpretable Image Recognitioncited by 63
  20. 2021How can I choose an explainer? An Application-grounded Evaluation of Post-hoc Explanationscited by 58
  21. 2020Towards falsifiable interpretability researchcited by 53
  22. 2022Post-Hoc Explanations Fail to Achieve their Purpose in Adversarial Contextscited by 53
  23. 2022Concept-level Debugging of Part-Prototype Networkscited by 52
  24. 2020This Looks Like That, Because ... Explaining Prototypes for Interpretable Image Recognitioncited by 50
  25. 2023FunnyBirds: A Synthetic Vision Dataset for a Part-Based Analysis of Explainable AI Methodscited by 49
  26. 2021This looks more like that: Enhancing Self-Explaining Models by Prototypical Relevance Propagationcited by 43
  27. 2023Learning Support and Trivial Prototypes for Interpretable Image Classificationcited by 38
  28. 2019Can I Trust the Explainer? Verifying Post-hoc Explanatory Methodscited by 37
  29. 2022Evaluation and Improvement of Interpretability for Self-Explainable Part-Prototype Networkscited by 37
  30. 2022ProtoVAE: A Trustworthy Self-Explainable Prototypical Variational Modelcited by 34
  31. 2023ProtoSeg: Interpretable Semantic Segmentation with Prototypical Partscited by 33
  32. 2023Interpretability Benchmark for Evaluating Spatial Misalignment of Prototypical Parts Explanationscited by 27
  33. 2022Proto2Proto: Can you recognize the car, the way I do?cited by 23
  34. 2019The Receptive Field as a Regularizer in Deep Convolutional Neural Networks for Acoustic Scene Classificationcited by 22
  35. 2022Towards Prototype-Based Self-Explainable Graph Neural Networkcited by 14
  36. 2022Unfooling Perturbation-Based Post Hoc Explainerscited by 12
  37. 2017Generalizing the Convolution Operator in Convolutional Neural Networkscited by 9
  38. 2022ML Interpretability: Simple Isn't Easycited by 7

50 most cited of 52 All 52 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 7; 2025: 7; 2026: 3.

18 papers.

Sort:

  1. 2024Interpretable Image Classification with Adaptive Prototype-based Vision Transformerscited by 25
  2. 2025A Comprehensive Survey on Self-Interpretable Neural Networkscited by 18
  3. 2024LucidPPN: Unambiguous Prototypical Parts Network for User-centric Interpretable Computer Visioncited by 15
  4. 2023Mixture of Gaussian-distributed Prototypes with Generative Modelling for Interpretable and Trustworthy Image Recognitioncited by 11
  5. 2025Interpretable Image Classification via Non-parametric Part Prototype Learningcited by 11
  6. 2024ProtoS-ViT: Visual foundation models for sparse self-explainable classificationscited by 8
  7. 2024Post-hoc Part-prototype Networkscited by 7
  8. 2024Applications of interpretable deep learning in neuroimaging: a comprehensive reviewcited by 3
  9. 2024This Probably Looks Exactly Like That: An Invertible Prototypical Networkcited by 3
  10. 2025This looks like what? Challenges and Future Research Directions for Part-Prototype Modelscited by 3
  11. 2025Tell me why: Visual foundation models as self-explainable classifierscited by 2
  12. 2025Debugging Concept Bottleneck Models through Removal and Retrainingcited by 2
6 more
  1. 2024Enhancing Interpretability of Vertebrae Fracture Grading using Human-interpretable Prototypescited by 0
  2. 2025Restricted Receptive Fields for Face Verificationcited by 0
  3. 2025Keypoint Counting Classifiers: Turning Vision Transformers into Self-Explainable Models Without Trainingcited by 0
  4. 2026ProtoQuant: Quantization of Prototypical Parts For General and Fine-Grained Image Classificationcited by 0
  5. 2026Diverse by Design: Architectural Constraints for Prototype-Based Interpretabilitycited by 0
  6. 2026Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbationscited by 0