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...

पूर्ण विवरण

ग्रंथसूची विवरण

मुख्य लेखकों:	Tengfei Shi, Shihai Wang, Bin Liu
स्वरूप:	Artigo
भाषा:	Inglês
प्रकाशित:	MDPI AG 2026-04-01
श्रृंखला:	Applied Sciences
विषय:	adversarial examples deep neural networks black-box adversarial attack transferable attack iterative attacks transferability
ऑनलाइन पहुंच:	https://www.mdpi.com/2076-3417/16/8/3748
टैग:	टैग जोड़ें कोई टैग नहीं, इस रिकॉर्ड को टैग करने वाले पहले व्यक्ति बनें!