A Survey of Explainable AI in Deep Visual Modeling: Methods and Metrics
This foundational survey addresses the critical need for interpretability in deep visual models used in high-stakes domains, though it is an incremental contribution as a review rather than novel research.
This paper provides the first comprehensive survey of explainable AI methods and evaluation metrics specifically for deep visual models, organizing existing techniques taxonomically and collating metrics to measure different explanation properties.
Deep visual models have widespread applications in high-stake domains. Hence, their black-box nature is currently attracting a large interest of the research community. We present the first survey in Explainable AI that focuses on the methods and metrics for interpreting deep visual models. Covering the landmark contributions along the state-of-the-art, we not only provide a taxonomic organization of the existing techniques, but also excavate a range of evaluation metrics and collate them as measures of different properties of model explanations. Along the insightful discussion on the current trends, we also discuss the challenges and future avenues for this research direction.