Selected papers and possible corresponding codes for part-prototype-based papers for XAI.

If you find a paper missed in this repository, please feel free to pull a request here.

Prototype-based methods are inclined to achieve interpretable image classification, which dissect the image by finding prototypical parts, and combine evidence from the prototypes to make a final classification. Each prototype is a feature vector that corresponds to a specific part of the object. With these prototypes, these methods can explain the reasoning process of a deep model like human beings, and they achieve comparable or even better accuracy with their analogous non-interpretable counterpart.

• Towards Interpretable Deep Reinforcement Learning with Human-Friendly Prototypes (ICLR 2023)

  • [Paper]

• ProtoVAE: A Trustworthy Self-Explainable Prototypical Variational Model (NIPS 2022)

  • [Paper]

• Semi-ProtoPNet Deep Neural Network for the Classification of Defective Power Grid Distribution Structures (Sensors 2022)

  • [Paper]

• Knowledge Distillation to Ensemble Global and Interpretable Prototype-Based Mammogram Classification Models (MICCAI 2022)

  • [Paper]

• ProtoPool: Interpretable Image Classification with Differentiable Prototypes Assignment (ECCV 2022)

  • [Paper] [Code]

• ViT-NeT: Interpretable Vision Transformers with Neural Tree Decoder (ICML 2022)

  • [Paper] [Code]

• Deformable ProtoPNet: An Interpretable Image Classifier Using Deformable Prototypes (CVPR 2022)

  • [Paper] [Code]

• Proto2Proto: Can you recognize the car, the way I do? (CVPR 2022)

  • [Paper] [Code]

• ProtGNN: Towards Self-Explaining Graph Neural Networks (AAAI 2022)

  • [Paper] [Code]

• ProtoMIL: Multiple Instance Learning with Prototypical Parts for Whole-Slide Image Classification (ECML PKDD 2022)

  • [Paper] [Code]

• ProtoPFormer: Concentrating on Prototypical Parts in Vision Transformers for Interpretable Image Recognition (Arxiv)

  • [Paper] [Code]

• Concept-level Debugging of Part-Prototype Networks (Arxiv)

  • [Paper] [Code]

• Extending explainability of generative classifiers with prototypical parts (Essay.Twente)

  • [Paper] [Code]

• But that's not why: Inference adjustment by interactive prototype deselection (Arxiv)

  • [Paper]

• Learnable Visual Words for Interpretable Image Recognition (Arxiv)

  • [Paper] [Code]

• Multi-Grained Interpretable Network for Image Recognition

  • [Paper]

• Neural Prototype Trees for Interpretable Fine-grained Image Recognition (CVPR 2021)

  • [Paper] [Code]

• ProtoPShare: Prototypical Parts Sharing for Similarity Discovery in Interpretable Image Classification (KDD 2021)

  • [Paper] [Code]

• Interpretable Image Recognition by Constructing Transparent Embedding Space (ICCV 2021)

  • [Paper] [Code]

• These do not Look Like Those: An Interpretable Deep Learning Model for Image Recognition (IEEE Access)

  • [Paper]

• Leveraging Class Hierarchies with Metric-Guided Prototype Learning (BMVC 2021)

  • [Paper] [Code]

• This looks like that, because... explaining prototypes for interpretable image recognition (Arxiv)

  • [Paper] [Code]

• This Looks Like That... Does it? Shortcomings of Latent Space Prototype Interpretability in Deep Networks (Arxiv)

  • [Paper] [Code]

• This Looks Like That: Deep Learning for Interpretable Image Recognition (NIPS 2019)

  • [Paper] [Code]

• Interpretable Image Recognition with Hierarchical Prototypes (HCOMP 2019)

  • [Paper] [Code]

• Deep Learning for Case-Based Reasoning through Prototypes: A Neural Network that Explains Its Predictions (AAAI 2018)
  • [Paper] [Code]