SifterNet uses Hopfield associative memory, trained on clean seed images, to purify backdoor triggers from poisoned inputs without accessing the target model.
Differential analysis of triggers and benign features for black-box DNN backdoor detection.IEEE Trans
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.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
SifterNet: A Generalized and Model-Agnostic Trigger Purification Approach
SifterNet uses Hopfield associative memory, trained on clean seed images, to purify backdoor triggers from poisoned inputs without accessing the target model.