Faster Algorithms for User-Level Private Stochastic Convex Optimization
We study private stochastic convex optimization (SCO) under user-level differential privacy (DP) constraints. In this scenario, there is north<annotation encoding="application/x-tex">northnorth ...