Abstract: We present a form of dataset augmentation for microwave imaging deep learning schemes that leverages the relationship between the assumed background medium and the physics of the underlying ...