arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2209.07403cs.LGcs.CRmath.OCstat.ML
收藏

具有大最坏情况Lipschitz参数的私有随机优化

Private Stochastic Optimization With Large Worst-Case Lipschitz Parameter

University of Southern California(南加州大学)

Andrew Lowy, Meisam Razaviyayn

AI总结:

本文针对损失函数最坏情况Lipschitz参数极大或无穷的差分隐私随机优化问题,提出基于随机梯度k阶矩的风险界和线性时间算法,覆盖光滑凸、非光滑凸及满足Proximal-PL不等式的非凸情形,并证明了部分结果的极小极大最优性。

摘要:

我们研究差分隐私(DP)随机优化(SO),其损失函数在所有数据上的最坏情况Lipschitz参数可能极大或无穷。迄今为止,绝大多数关于DP SO的工作假设损失函数在数据上是一致Lipschitz连续的(即随机梯度一致有界)。虽然这一假设方便,但它往往导致悲观的风险界。在许多实际问题中,由于离群值和/或重尾数据,损失函数在所有数据上的最坏情况(一致)Lipschitz参数可能非常巨大。在这种情况下,随最坏情况Lipschitz参数缩放的DP SO风险界是空洞无意义的。为了解决这些局限性,我们提供了不依赖于一致Lipschitz参数的改进风险界。遵循最近的一系列工作[WXDX20, KLZ22],我们假设随机梯度对某个$k \geq 2$具有有界的$k$阶矩。与一致Lipschitz DP SO的工作相比,我们的风险界随损失的$k$阶矩而非一致Lipschitz参数缩放,从而在存在离群值和/或重尾数据时允许显著更快的收敛速率。对于光滑凸损失函数,我们提供了具有最先进超额风险的线性时间算法。我们用新颖的下界补充了我们的超额风险上界。在某些参数区间内,我们的线性时间超额风险界是极小极大最优的。其次,我们提供了第一个处理非光滑凸损失函数的算法。为此,我们开发了新颖的基于算法和稳定性的证明技术,我们相信这些技术将对未来获得最优超额风险的工作有用。最后,我们的工作是第一个处理满足Proximal-PL不等式的非凸非一致Lipschitz损失函数的;这涵盖了一些实用的机器学习模型。我们的Proximal-PL算法具有近最优的超额风险。

英文摘要:

We study differentially private (DP) stochastic optimization (SO) with loss functions whose worst-case Lipschitz parameter over all data may be extremely large or infinite. To date, the vast majority of work on DP SO assumes that the loss is uniformly Lipschitz continuous (i.e. stochastic gradients are uniformly bounded) over data. While this assumption is convenient, it often leads to pessimistic risk bounds. In many practical problems, the worst-case (uniform) Lipschitz parameter of the loss over all data may be huge due to outliers and/or heavy-tailed data. In such cases, the risk bounds for DP SO, which scale with the worst-case Lipschitz parameter, are vacuous. To address these limitations, we provide improved risk bounds that do not depend on the uniform Lipschitz parameter. Following a recent line of work [WXDX20, KLZ22], we assume that stochastic gradients have bounded $k$-th order moments for some $k \geq 2$. Compared with works on uniformly Lipschitz DP SO, our risk bounds scale with the $k$-th moment instead of the uniform Lipschitz parameter of the loss, allowing for significantly faster rates in the presence of outliers and/or heavy-tailed data. For smooth convex loss functions, we provide linear-time algorithms with state-of-the-art excess risk. We complement our excess risk upper bounds with novel lower bounds. In certain parameter regimes, our linear-time excess risk bounds are minimax optimal. Second, we provide the first algorithm to handle non-smooth convex loss functions. To do so, we develop novel algorithmic and stability-based proof techniques, which we believe will be useful for future work in obtaining optimal excess risk. Finally, our work is the first to address non-convex non-uniformly Lipschitz loss functions satisfying the Proximal-PL inequality; this covers some practical machine learning models. Our Proximal-PL algorithm has near-optimal excess risk.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑