A new black-box adversarial example detector, PID, uses the confidence of an auxiliary model on the primary model's predicted label (1 - g_y(x)) to separate adversarial from normal inputs.
Fast is better than free: Revisiting adversarial training
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
method 1
citation-polarity summary
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1roles
method 1polarities
use method 1representative citing papers
citing papers explorer
-
Prediction Inconsistency Helps Achieve Generalizable Detection of Adversarial Examples
A new black-box adversarial example detector, PID, uses the confidence of an auxiliary model on the primary model's predicted label (1 - g_y(x)) to separate adversarial from normal inputs.