Hanbin Hong
Hanbin Hong
Home
Publications
Projects
Contact
Light
Dark
Automatic
3
Certifiable Black-Box Attack: Ensuring Provably Successful Attack for Adversarial Examples
We propose the first black-box adversarial attack with the certifiable guarantee to the attack success rate.
Hanbin Hong
,
Yuan Hong
Cite
Certified Adversarial Robustness via Anisotropic Randomized Smoothing
We propose the first anisotropic randomized smoothing method which ensures provable robustness guarantee based on pixel-wise noise distributions.
Hanbin Hong
,
Yuan Hong
PDF
Cite
Cite
×