Improving Transferability of Adversarial Attacks via Frequency-Consistent Regularization

Adversarial examples have revealed the vulnerability of deep neural networks, and their transferability makes black-box attacks particularly concerning. However, perturbations crafted on a surrogate model often do not remain sufficiently effective on unseen target models. In this paper, we revisit t...

Celý popis

Podrobná bibliografie

Hlavní autoři:	Tengfei Shi, Shihai Wang, Bin Liu
Médium:	Artigo
Jazyk:	Inglês
Vydáno:	MDPI AG 2026-04-01
Edice:	Applied Sciences
Témata:	adversarial examples deep neural networks black-box adversarial attack transferable attack iterative attacks transferability
On-line přístup:	https://www.mdpi.com/2076-3417/16/8/3748
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!