ProtoPShare: Prototype Sharing for Interpretable Image Classification and Similarity Discovery

Abstract

In this paper, we introduce ProtoPShare, a self-explained method that incorporates the paradigm of prototypical parts to explain its predictions. The main novelty of the ProtoPShare is its ability to efficiently share prototypical parts between the classes thanks to our data-dependent merge-pruning. Moreover, the prototypes are more consistent and the model is more robust to image perturbations than the state of the art method ProtoPNet. We verify our findings on two datasets, the CUB-200-2011 and the Stanford Cars.

References in corpus (6)

• Interpretable Explanations of Black Boxes by Meaningful Perturbation
• Sanity Checks for Saliency Maps
• Channel Pruning for Accelerating Very Deep Neural Networks
• Rethinking the Smaller-Norm-Less-Informative Assumption in Channel Pruning of Convolution Layers
• Interpretable and Steerable Sequence Learning via Prototypes
• Exploring Linear Relationship in Feature Map Subspace for ConvNets Compression

Cited by in corpus (8)

• Explainable Artificial Intelligence (XAI) 2.0: A Manifesto of Open Challenges and Interdisciplinary Research Directions
• A Comprehensive Survey on Self-Interpretable Neural Networks
• DP-Net: Learning Discriminative Parts for image recognition
• Multi-Scale Grouped Prototypes for Interpretable Semantic Segmentation
• Comprehensive Evaluation of Prototype Neural Networks
• ProtoP-OD: Explainable Object Detection with Prototypical Parts
• Birds look like cars: Adversarial analysis of intrinsically interpretable deep learning
• Visual Probing: Cognitive Framework for Explaining Self-Supervised Image Representations