SD-MIA is a black-box membership inference attack that detects pre-training data in diffusion models via cross-modal perturbations on images and textual instructions.
Real-world benchmarks make membership inference attacks fail on diffusion models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Black-box Membership Inference Attacks on the Pre-training Data of Image-generation Models
SD-MIA is a black-box membership inference attack that detects pre-training data in diffusion models via cross-modal perturbations on images and textual instructions.