Towards Desiderata-Driven Design of Visual Counterfactual Explainers
This work addresses the need for more comprehensive explanations in image classifiers to enhance transparency for users and developers, though it is incremental as it builds on existing VCE methods.
The paper tackles the problem that existing visual counterfactual explainers (VCEs) focus too narrowly on sample quality or change minimality, neglecting holistic desiderata like fidelity, understandability, and sufficiency, and proposes a novel 'smooth counterfactual explorer' (SCE) algorithm, demonstrating its effectiveness through evaluations on synthetic and real data.
Visual counterfactual explainers (VCEs) are a straightforward and promising approach to enhancing the transparency of image classifiers. VCEs complement other types of explanations, such as feature attribution, by revealing the specific data transformations to which a machine learning model responds most strongly. In this paper, we argue that existing VCEs focus too narrowly on optimizing sample quality or change minimality; they fail to consider the more holistic desiderata for an explanation, such as fidelity, understandability, and sufficiency. To address this shortcoming, we explore new mechanisms for counterfactual generation and investigate how they can help fulfill these desiderata. We combine these mechanisms into a novel 'smooth counterfactual explorer' (SCE) algorithm and demonstrate its effectiveness through systematic evaluations on synthetic and real data.