arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.12433math.STcs.DScs.LGstat.MLstat.TH
收藏

无方差的最优均值估计

Optimal Mean Estimation without a Variance

Department of Electrical Engineering and Computer Science(电气工程与计算机科学系);Department of Statistics(统计系)

Yeshwanth Cherapanamjeri, Nilesh Tripuraneni, Peter L. Bartlett, Michael I. Jordan

摘要:

我们研究的重尾均值估计的数据生成分布的方差不存在的设置中的问题。具体地说,给定一个样本$\mathbf{X} = \{X_i\}_{i = 1}^n$从一个分布$\mathcal{D}$在$\mathbb{R}^d$与平均$μ$满足以下\emph{weak-moment}假设的一些${α\in [0, 1]}$:\begin{equation*} \forall \|v\| = 1: \mathbb{E}_{X \thicksim \mathcal{D}}[\lvert \langle X - μ, v\rangle \rvert^{1 + α}] \leq 1, \end{equation*}和给定的目标故障概率,$δ$,我们的目标是设计一个估计,达到最小可能的置信区间作为$n,d,δ$的函数。对于$α= 1$的特定情况,Lugosi和Mendelson的基础工作展示了一个估计器,该估计器实现了亚高斯置信区间,随后的工作导致了该估计器的计算效率版本。在这里,我们研究的情况下,一般$α$,并建立以下信息理论的最佳可达到的置信区间的下限:\begin{equation*} Ω\left(\sqrt{\frac{d}{n}} + \left(\frac{d}{n}\right)^{\fracα{(1 + α)}} + \left(\frac{\log 1 / δ}{n}\right)^{\fracα{(1 + α)}}\right). \end{equation*}此外,我们设计了一个计算效率的估计,达到这个下限。

英文摘要:

We study the problem of heavy-tailed mean estimation in settings where the variance of the data-generating distribution does not exist. Concretely, given a sample $\mathbf{X} = \{X_i\}_{i = 1}^n$ from a distribution $\mathcal{D}$ over $\mathbb{R}^d$ with mean $μ$ which satisfies the following \emph{weak-moment} assumption for some ${α\in [0, 1]}$: \begin{equation*} \forall \|v\| = 1: \mathbb{E}_{X \thicksim \mathcal{D}}[\lvert \langle X - μ, v\rangle \rvert^{1 + α}] \leq 1, \end{equation*} and given a target failure probability, $δ$, our goal is to design an estimator which attains the smallest possible confidence interval as a function of $n,d,δ$. For the specific case of $α= 1$, foundational work of Lugosi and Mendelson exhibits an estimator achieving subgaussian confidence intervals, and subsequent work has led to computationally efficient versions of this estimator. Here, we study the case of general $α$, and establish the following information-theoretic lower bound on the optimal attainable confidence interval: \begin{equation*} Ω\left(\sqrt{\frac{d}{n}} + \left(\frac{d}{n}\right)^{\fracα{(1 + α)}} + \left(\frac{\log 1 / δ}{n}\right)^{\fracα{(1 + α)}}\right). \end{equation*} Moreover, we devise a computationally-efficient estimator which achieves this lower bound.

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。

↑