Alike Parts: A Feature-Informed Approach to Local and Global Prototype Explanations

2026-05-20

AI Summary

This work addresses the limited feature-level granularity of existing prototype-based explanation methods, which often fail to reveal the precise basis of model decisions. It proposes a novel framework that integrates feature importance into both local explanations and global prototype selection. Locally, the method highlights key features shared between an instance and its nearest prototype through “similar parts,” thereby clarifying decision rationales. Globally, feature importance is incorporated into the prototype selection objective to enhance feature diversity among prototypes. The approach combines feature importance scoring, nearest-neighbor prototype matching, and a diversity-constrained optimization objective. Experiments across six benchmark datasets demonstrate that the method significantly improves the interpretability and feature diversity of prototype explanations while maintaining or even enhancing the fidelity of surrogate model predictions.

Abstract

Prototype-based explanations offer an intuitive, example-based approach to support the interpretability of machine learning black box classifiers but often lack feature-level granularity. We introduce a framework that integrates feature importance at two levels to address this gap. First, for local explanations, we propose \textit{alike parts}: a method that uses feature importance scores to highlight the most relevant, shared feature subsets between a classified instance and its nearest prototype, guiding user attention. Second, we augment the global prototype selection objective function with a feature importance term to actively promote diversity in the feature attributions of the selected prototypes. Experiments on six benchmark datasets show that this augmented selection process maintains or, in some cases, increases the prediction fidelity of the surrogate model, suggesting that feature diversity does not compromise model fidelity.

Problem

• prototype explanations
• feature importance
• interpretability
• black box classifiers
• feature-level granularity

Innovation

• prototype-based explanation
• feature importance
• alike parts
• interpretability
• surrogate model