Causal Explanations for Image Classifiers

TL;DR

This paper introduces a principled, causality-based approach for explaining image classifier outputs, providing a new algorithm, theoretical guarantees, and demonstrating superior efficiency and explanation quality compared to existing tools.

Contribution

The paper presents a novel black-box explanation method grounded in actual causality theory, with proven termination, efficiency, and improved explanation quality.

Findings

ReX is the most efficient black-box explanation tool.

ReX produces the smallest explanations among compared methods.

ReX outperforms state-of-the-art tools on standard quality metrics.

Abstract

Existing algorithms for explaining the output of image classifiers use different definitions of explanations and a variety of techniques to find them. However, none of the existing tools use a principled approach based on formal definitions of cause and explanation. In this paper we present a novel black-box approach to computing explanations grounded in the theory of actual causality. We prove relevant theoretical results and present an algorithm for computing approximate explanations based on these definitions. We prove termination of our algorithm and discuss its complexity and the amount of approximation compared to the precise definition. We implemented the framework in a tool ReX and we present experimental results and a comparison with state-of-the-art tools. We demonstrate that ReX is the most efficient black-box tool and produces the smallest explanations, in addition to outperforming other black-box tools on standard quality measures.