Concept-based Adversarial Attacks: Tricking Humans and Classifiers Alike

Workshop 43th IEEE Symposium on Security and Privacy -- 5th Deep Learning Security Workshop

San Francisco, California, USA

Oneliner: The only presentation done physically at [DLS22]!

This was the first presentation I gave in person since the beginning of the Covid19 pandemic. The last one was the one for NCA19 (September 2019), and this one was in May 2022: a whopping 32 months without physical meetings.

When I received the notification that my paper was accepted at DLS22, I knew that nothing would have stopped me from going to Oakland in person. Despite being awake for more than 60 hours, and despite having some “problems” at the security checks in Zurich, I was finally able to put an end to the curse of “remote meetings”.

Paper Slides Video (TBD) Venue