A clean-label targeted poisoning method that matches gradient magnitude as well as direction generalizes to unseen physical variations of a target object, reaching 90.13% success on multi-view cars.
Ronny Huang, Wojciech Czaja, Gavin Taylor, Michael Moeller, and Tom Goldstein
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
-
Generalizable Targeted Data Poisoning against Varying Physical Objects
A clean-label targeted poisoning method that matches gradient magnitude as well as direction generalizes to unseen physical variations of a target object, reaching 90.13% success on multi-view cars.