Extracting Explanations, Justification, and Uncertainty from Black-Box Deep Neural Networks

Efficiently extract explanations, justifications, and uncertainty from black-box DNNs using a novel Bayesian approach.