Causal Explanations for Image Classifiers
This provides a principled method for generating causal explanations in image classification, addressing a key interpretability challenge for users of AI systems, though it is incremental in applying formal causality to an existing problem.
The paper tackles the problem of explaining image classifier outputs by introducing a black-box approach grounded in actual causality theory, resulting in a tool (ReX) that is the most efficient, produces the smallest explanations, and outperforms other black-box tools on standard quality measures.
Existing algorithms for explaining the output of image classifiers use different definitions of explanations and a variety of techniques to extract them. However, none of the existing tools use a principled approach based on formal definitions of causes and explanations for the explanation extraction. 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 tool and produces the smallest explanations, in addition to outperforming other black-box tools on standard quality measures.