A local-training algorithm for nonconvex distributed optimization achieves communication efficiency and differential privacy via gradient clipping plus additive noise, with proven convergence to a stationary point within bounded distance and formal privacy guarantees.
Deep learning with differential privacy
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
support 1representative citing papers
SnapAudit decomposes DP-ICL into a deterministic snapshot stage and a stochastic noise stage, using bootstrap simulation to achieve 80-200x faster auditing and exposing privacy bound violations in existing Gaussian and embedding mechanisms.
citing papers explorer
-
Communication-Efficient Distributed Learning with Differential Privacy
A local-training algorithm for nonconvex distributed optimization achieves communication efficiency and differential privacy via gradient clipping plus additive noise, with proven convergence to a stationary point within bounded distance and formal privacy guarantees.
-
SnapAudit: Active Auditing of Differentially Private In-Context Learning via Snapshot-Based Simulation
SnapAudit decomposes DP-ICL into a deterministic snapshot stage and a stochastic noise stage, using bootstrap simulation to achieve 80-200x faster auditing and exposing privacy bound violations in existing Gaussian and embedding mechanisms.