arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2312.02417math.STcs.DSstat.MLstat.TH
收藏

未知异方差情形下的近最优均值估计

Near-Optimal Mean Estimation with Unknown, Heteroskedastic Variances

Spencer Compton, Gregory Valiant

AI总结:

本文针对未知异方差高斯数据的共同均值估计问题,提出一种直观高效且在信号子集模型下达到近最优误差的算法,将所需方差有界样本数从 Ω̃(n^{1/2}) 降至 Õ(n^{1/4}),并推广到多维情形。

摘要:

给定来自一组具有共同均值但方差不同且未知的高斯变量的数据,估计其共同均值的最佳算法是什么?我们针对该任务提出了一种直观且高效的算法。由于不同的闭式保证难以比较,信号子集模型可作为异方差均值估计的基准:给定 n 个高斯变量,其中未知的 m 个变量方差以 1 为界,那么作为 n 和 m 的函数,最优估计误差是多少?我们的算法在相差对数因子的意义上解决了这一开放问题,当 m = n^c(对所有 0<c<1)时,将此前已知的最优估计误差改进了多项式因子。特别值得注意的是,我们仅用 m = Õ(n^{1/4}) 个方差有界样本就获得了 o(1) 的误差,而此前的工作需要 m = Ω̃(n^{1/2})。最后,我们证明在多维情形下,即使 d=2,我们的技术也能达到与已知每个样本方差时相当的收敛速率。

英文摘要:

Given data drawn from a collection of Gaussian variables with a common mean but different and unknown variances, what is the best algorithm for estimating their common mean? We present an intuitive and efficient algorithm for this task. As different closed-form guarantees can be hard to compare, the Subset-of-Signals model serves as a benchmark for heteroskedastic mean estimation: given $n$ Gaussian variables with an unknown subset of $m$ variables having variance bounded by 1, what is the optimal estimation error as a function of $n$ and $m$? Our algorithm resolves this open question up to logarithmic factors, improving upon the previous best known estimation error by polynomial factors when $m = n^c$ for all $0<c<1$. Of particular note, we obtain error $o(1)$ with $m = \tilde{O}(n^{1/4})$ variance-bounded samples, whereas previous work required $m = \tildeΩ(n^{1/2})$. Finally, we show that in the multi-dimensional setting, even for $d=2$, our techniques enable rates comparable to knowing the variance of each sample.

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。

↑