CARE replaces the raw target direction in value-space concept erasure with a covariance-aware retained-subspace direction, reducing collateral damage to non-target concepts at negligible computational cost.
In: International Con- ference on Learning Representations (2024) 3
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Erasing Without Collateral Damage: Precise Concept Removal in Diffusion Models
CARE replaces the raw target direction in value-space concept erasure with a covariance-aware retained-subspace direction, reducing collateral damage to non-target concepts at negligible computational cost.