Convergent Privacy Loss o
Convergent Privacy Loss of Noisy-SGD without Convexity and Smoothness
Convergent Privacy Loss of Noisy-SGD without Convexity and Smoothness
arXiv:2410.01068v2 Announce Type: replace
Abstract: We study the Differential Privacy (DP) guarantee of hidden-state Noisy-SGD algorithms over a bounded domain. Standard privacy analysis for Noisy-SGD assumes all internal states are revealed, which leads to a divergent R’enyi DP bound with respect …
