SentiNet: Detecting Localized Universal Attacks Against Deep Learning Systems

Publication
Arxiv