Exiaa: Explainable Injections for Adversarial Attack

Leonardo Pesce, Jiawen Wei, Gianmarco Mengaldo

Submission Type: I want my submission to be considered for both oral and poster presentation.

Keywords: Explainable AI, Machine Learning, Interpretability, Adversarial attack

TL;DR: We introduce a novel non-iterative model agnostic black-box attack to explanations by exploiting an intrinsic flaw of neural networks and post-hoc explainable methods.

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 311