mlsanityData to 13 Oct 2026

Improving Prototypical Visual Explanations with Reward Reweighing, Reselection, and Retraining

arXiv 2307.03887 ↗First posted 8 Jul 2023cs.LG cs.AI cs.CVcited by 3

On the map: LLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment

Abstract (arXiv)

In recent years, work has gone into developing deep interpretable methods for image classification that clearly attributes a model's output to specific features of the data. One such of these methods is the Prototypical Part Network (ProtoPNet), which attempts to classify images based on meaningful parts of the input. While this architecture is able to produce visually interpretable classifications, it often learns to classify based on parts of the image that are not semantically meaningful. To address this problem, we propose the Reward Reweighing, Reselecting, and Retraining (R3) post-processing framework, which performs three additional corrective updates to a pretrained ProtoPNet in an offline and efficient manner. The first two steps involve learning a reward model based on collected human feedback and then aligning the prototypes with human preferences. The final step is retraining, which realigns the base features and the classifier layer of the original model with the updated prototypes. We find that our R3 framework consistently improves both the interpretability and the predictive accuracy of ProtoPNet and its variants.

Key facts. Cited by 3 papers in the map (data through 13 Oct 2026). Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 24 of its 32 references that are in the map. First posted on arXiv 8 Jul 2023.

Builds on

24 of 32 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Aligning Text-to-Image Models using Human Feedback23 Feb

Other references · 23

  1. 2017A Unified Approach to Interpreting Model Predictionscited by 6,515
  2. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,055
  3. 2017Deep reinforcement learning from human preferencescited by 3,871
  4. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,691
  5. 2019Fine-Tuning Language Models from Human Preferencescited by 2,139
  6. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,410
  7. 2021A General Language Assistant as a Laboratory for Alignmentcited by 1,041
  8. 2018This Looks Like That: Deep Learning for Interpretable Image Recognitioncited by 735
  9. 2021Interpretable Machine Learning: Fundamental Principles and 10 Grand Challengescited by 365
  10. 2019Interpretable machine learning: definitions, methods, and applicationscited by 305
  11. 2020Reward-rational (implicit) choice: A unifying formalism for reward learningcited by 120
  12. 2021Deformable ProtoPNet: An Interpretable Image Classifier Using Deformable Prototypescited by 120
11 more
  1. 2020ProtoPShare: Prototype Sharing for Interpretable Image Classification and Similarity Discoverycited by 89
  2. 2021IAIA-BL: A Case-based Interpretable Deep Learning Model for Classification of Mass Lesions in Digital Mammographycited by 72
  3. 2018Beyond Word Importance: Contextual Decomposition to Extract Interactions from LSTMscited by 67
  4. 2019Veridical Data Sciencecited by 65
  5. 2022ProtoPFormer: Concentrating on Prototypical Parts in Vision Transformers for Interpretable Image Recognitioncited by 63
  6. 2019Towards Explainable Deep Neural Networks (xDNN)cited by 61
  7. 2020This Looks Like That, Because ... Explaining Prototypes for Interpretable Image Recognitioncited by 50
  8. 2022Evaluation and Improvement of Interpretability for Self-Explainable Part-Prototype Networkscited by 37
  9. 2018Human-in-the-Loop Interpretability Priorcited by 35
  10. 2021Self-Interpretable Model with TransformationEquivariant Interpretationcited by 21
  11. 2022Hierarchical Shrinkage: improving the accuracy and interpretability of tree-based methodscited by 18

Built on by

Cited by year

Citing papers per year: 2025: 3.

3 papers.

  1. 2025Rashomon Sets for Prototypical-Part Networks: Editing Interpretable Models in Real-Timecited by 13
  2. 2025This looks like what? Challenges and Future Research Directions for Part-Prototype Modelscited by 3
  3. 2025Adaptive Test-Time Intervention for Concept Bottleneck Modelscited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2025Interpretable Image Classification via Non-parametric Part Prototype Learning
  2. 2023This Looks Like Those: Illuminating Prototypical Concepts Using Multiple Visualizations
  3. 2026ProtoQuant: Quantization of Prototypical Parts For General and Fine-Grained Image Classification
  4. 2022Towards Human-Interpretable Prototypes for Visual Assessment of Image Classification Models
  5. 2022Prompting through Prototype: A Prototype-based Prompt Learning on Pretrained Vision-Language Models
  6. 2026Reprogramming Vision-Language Models via Structured Prompt Reparameterization
  7. 2023Proto-CLIP: Vision-Language Prototypical Network for Few-Shot Learning
  8. 2026Supervised Classification Heads as Semantic Prototypes: Unlocking Vision-Language Alignment via Weight Recycling