Differential privacy in policy optimization adds sample complexity costs that often appear as lower-order terms rather than dominating the bounds.
Private (stochastic) non-convex optimization revisited: Second-order stationary points and excess risks.Advances in Neural Information Processing Systems, 36:65618–65641, 2023
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Sample Complexity of Differentially Private Policy Optimization
Differential privacy in policy optimization adds sample complexity costs that often appear as lower-order terms rather than dominating the bounds.