Machine learningOther

Cutting through buggy adversarial example defenses: fixing 1 line of code breaks Sabre

Buggy Defenses: The paper exposes major flaws in the Sabre defense against adversarial examples, including gradient masking and bugs in the evaluation code, which when corrected, drop Sabre's robust accuracy to 0%.