Multivariate mean estimation with direction-dependent accuracy

DOI10.4171/JEMS/1321arXiv2010.11921OpenAlexW3094564758MaRDI QIDQ6352013FDOQ6352013

Authors: Gábor Lugosi, Shahar Mendelson 

Publication date: 22 October 2020

Abstract: We consider the problem of estimating the mean of a random vector based on $N$ independent, identically distributed observations. We prove the existence of an estimator that has a near-optimal error in all directions in which the variance of the one dimensional marginal of the random vector is not too small: with probability $1-delta$, the procedure returns $wh{mu}_N$ which satisfies that for every direction $uin{S}^{d-1}$, [ inr{wh{mu}_N - mu, u}le frac{C}{sqrt{N}} left( sigma(u)sqrt{log(1/delta)} + left(E|X-EXP X|_2^2 ight)^{1/2} ight)~, ] where $sigm{a}^2(u)=var(inrX,u)$ and $C$ is a constant. To achieve this, we require only slightly more than the existence of the covariance matrix, in the form of a certain moment-equivalence assumption. The proof relies on novel bounds for the ratio of empirical and true probabilities that hold uniformly over certain classes of random variables.

Full work available at URL: https://doi.org/10.4171/jems/1321

Mathematics Subject Classification ID

Nonparametric estimation (62G05)

Cited In (2)

• The geometric median and applications to robust mean estimation
• Empirical approximation of the Gaussian distribution in \(\mathbb{R}^d\)

This page was built for publication: Multivariate mean estimation with direction-dependent accuracy

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6352013)