Multiple Different Black Box Explanations for Image Classifiers

TL;DR

This paper introduces MultEX, an algorithm that generates multiple high-quality explanations for image classifier decisions, providing deeper insights into the classifier's behavior beyond single explanations.

Contribution

The paper presents a novel algorithm, MultEX, based on actual causality, capable of producing multiple explanations for black-box image classifiers, enhancing interpretability.

Findings

MultEX finds more explanations than existing methods.

The explanations generated by MultEX are of higher quality.

MultEX performs well across different models and datasets.

Abstract

Existing explanation tools for image classifiers usually give only a single explanation for an image's classification. For many images, however, image classifiers accept more than one explanation for the image label. These explanations are useful for analyzing the decision process of the classifier and for detecting errors. Thus, restricting the number of explanations to just one severely limits insight into the behavior of the classifier. In this paper, we describe an algorithm and a tool, MultEX, for computing multiple explanations as the output of a black-box image classifier for a given image. Our algorithm uses a principled approach based on actual causality. We analyze its theoretical complexity and evaluate MultEX against the state-of-the-art across three different models and three different datasets. We find that MultEX finds more explanations and that these explanations are of higher quality.