Beyond Attribution: Unified Concept-Level Explanations

TL;DR

This paper introduces UnCLE, a unified framework that extends model-agnostic explanation methods to generate diverse, concept-based explanations like attributions, sufficient conditions, and counterfactuals across different modalities.

Contribution

UnCLE is a novel framework that bridges the gap between model-agnostic and concept-based explanations, enabling diverse explanation forms for various models and modalities.

Findings

UnCLE provides more faithful explanations than existing methods.

It offers richer explanation forms satisfying different user needs.

Effective across text, image, and multimodal models.

Abstract

There is an increasing need to integrate model-agnostic explanation techniques with concept-based approaches, as the former can explain models across different architectures while the latter makes explanations more faithful and understandable to end-users. However, existing concept-based model-agnostic explanation methods are limited in scope, mainly focusing on attribution-based explanations while neglecting diverse forms like sufficient conditions and counterfactuals, thus narrowing their utility. To bridge this gap, we propose a general framework UnCLE to elevate existing local model-agnostic techniques to provide concept-based explanations. Our key insight is that we can uniformly extend existing local model-agnostic methods to provide unified concept-based explanations with large pre-trained model perturbation. We have instantiated UnCLE to provide concept-based explanations in three forms: attributions, sufficient conditions, and counterfactuals, and applied it to popular text, image, and multimodal models. Our evaluation results demonstrate that UnCLE provides explanations more faithful than state-of-the-art concept-based explanation methods, and provides richer explanation forms that satisfy various user needs.