A new per-secret protection definition and an LP-based sampling algorithm that trains models with substantially lower noise than DP-SGD while bounding the posterior probability of secret reconstruction.
Reconstructing training data with informed adversaries
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Hush! Protecting Secrets During Model Training: An Indistinguishability Approach
A new per-secret protection definition and an LP-based sampling algorithm that trains models with substantially lower noise than DP-SGD while bounding the posterior probability of secret reconstruction.