Balls-and-Bins sampling for DP-SGD has a tight privacy analysis: as private as Poisson at large epsilon, with shuffle-comparable utility, verified by Monte Carlo accounting.
It's our loss: No privacy amplification for hidden state DP-SGD with non-convex loss
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
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Balls-and-Bins Sampling for DP-SGD
Balls-and-Bins sampling for DP-SGD has a tight privacy analysis: as private as Poisson at large epsilon, with shuffle-comparable utility, verified by Monte Carlo accounting.