Averaging adversarial examples over the fine-tuning trajectory, called AaF, improves targeted adversarial transferability across CNN and transformer victim models compared with endpoint-only fine-tuning.
Evading defenses to transferable adversarial examples by translation -invariant attacks,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Two Heads Are Better Than One: Averaging along Fine-Tuning to Improve Targeted Transferability
Averaging adversarial examples over the fine-tuning trajectory, called AaF, improves targeted adversarial transferability across CNN and transformer victim models compared with endpoint-only fine-tuning.