Abstract: To mitigate recent insidious backdoor attacks on deep learning models, advances have been made by the research community. Nonetheless, state-of-the-art defenses are either limited to ...