论文标题
差异私人在线到批处理，以平滑损失
Differentially Private Online-to-Batch for Smooth Losses
论文作者
论文摘要
我们开发了一种新的减少，该减少将遭受$ O（\ sqrt {t}）$遗憾的任何在线凸优化算法转换为$ε$ - divertial私有随机凸优化算法，并具有最佳的收敛速率$ \ tilde o（1/\ sqrt $} $ smoper线性时间，与经典的非私人“在线到批量”转换形成直接类比。通过将我们的技术应用于更高级的自适应在线算法，我们生成了自适应差异私人对应物，其收敛率取决于APRIORI未知方差或参数规范。
We develop a new reduction that converts any online convex optimization algorithm suffering $O(\sqrt{T})$ regret into an $ε$-differentially private stochastic convex optimization algorithm with the optimal convergence rate $\tilde O(1/\sqrt{T} + \sqrt{d}/εT)$ on smooth losses in linear time, forming a direct analogy to the classical non-private "online-to-batch" conversion. By applying our techniques to more advanced adaptive online algorithms, we produce adaptive differentially private counterparts whose convergence rates depend on apriori unknown variances or parameter norms.
