The dithered Gaussian mechanism is presented, an alternative to the discrete Gaussian mechanism for differential privacy that discretizes the private output rather than the noise distribution itself, and it is shown that cryptographically secure noise generation with reduced exposure to floating-point vulnerabilities c...
Nikita P. Kalinin, Rasmus Pagh· arXiv.org· 1 citation
In the low-memory, low-bandwidth regime, $\gamma$-BIFR improves RMSE, amplified RMSE, and private training performance, while yielding tighter theoretical guarantees for multi-participation error in multi-epoch training.
Nikita P. Kalinin, Aki Rehn, Joel Daniel Andersson et al.· arXiv.org· 1 citation
This work proposes a new algorithm for correlated-noise private training that correlates noise only with the immediately preceding iteration and cancels a tunable portion of it and shows that the computational overhead is minimal and improved accuracy over DP-SGD is demonstrated.
Nikita P. Kalinin, Ryan McKenna, Rasmus Pagh et al.· arXiv.org· 3 citations
This work improves the leading constant in the maximum squared error and the mean squared error, and proves an $\Omega(\epsilon^{-2}\log^3 n)$ lower bound for the class of factorizations whose matrices have entries in $\{0,1\}$, matching the upper-bound asymptotics for this class.
P. Arkhipov, Nikita P. Kalinin· arXiv.org· 2 citations· ⚡1
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
