Global interpretability is a vital requirement for image classification in many application domains ranging from medical diagnosis to autonomous driving, from both a legal perspective but also for improving the classifier’s performance or fairness. Existing interpretability methods mainly explain a model behavior by identifying salient image patches, which require manual efforts from users to make sense of, and also do not typically support model validation with questions that contain multiple concepts. In this paper, we introduce a scalable, crowdsourcing-based human-in-the-loop approach for global interpretability. Salient image areas identified by local interpretability methods are annotated with semantic concepts, which are then aggregated into a tabular representation of images to facilitate automatic statistical analysis of model behavior. We show that this approach can answer both interpretability needs for model validation and exploration, and provides semantically more diverse, informative, and relevant explanations while still allowing for scalable and cost-efficient execution.

The Web Conference is announcing latest news and developments biweekly or on a monthly basis. We respect The General Data Protection Regulation 2016/679.