{"title":"具有方向相关精度的多元均值估计","authors":"G. Lugosi, S. Mendelson","doi":"10.4171/jems/1321","DOIUrl":null,"url":null,"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 $u \\in S^{d-1}$, \\[ \\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\\right)^{1/2} \\right)~, \\] where $\\sigma^2(u) = \\var(\\inr{X,u})$ 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. \nThe proof relies on novel bounds for the ratio of empirical and true probabilities that hold uniformly over certain classes of random variables.","PeriodicalId":50003,"journal":{"name":"Journal of the European Mathematical Society","volume":"51 1","pages":""},"PeriodicalIF":2.5000,"publicationDate":"2020-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":"{\"title\":\"Multivariate mean estimation with direction-dependent accuracy\",\"authors\":\"G. Lugosi, S. Mendelson\",\"doi\":\"10.4171/jems/1321\",\"DOIUrl\":null,\"url\":null,\"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 $u \\\\in S^{d-1}$, \\\\[ \\\\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\\\\right)^{1/2} \\\\right)~, \\\\] where $\\\\sigma^2(u) = \\\\var(\\\\inr{X,u})$ 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. \\nThe proof relies on novel bounds for the ratio of empirical and true probabilities that hold uniformly over certain classes of random variables.\",\"PeriodicalId\":50003,\"journal\":{\"name\":\"Journal of the European Mathematical Society\",\"volume\":\"51 1\",\"pages\":\"\"},\"PeriodicalIF\":2.5000,\"publicationDate\":\"2020-10-22\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"5\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of the European Mathematical Society\",\"FirstCategoryId\":\"100\",\"ListUrlMain\":\"https://doi.org/10.4171/jems/1321\",\"RegionNum\":1,\"RegionCategory\":\"数学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"MATHEMATICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of the European Mathematical Society","FirstCategoryId":"100","ListUrlMain":"https://doi.org/10.4171/jems/1321","RegionNum":1,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATHEMATICS","Score":null,"Total":0}
Multivariate mean estimation with direction-dependent accuracy
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 $u \in S^{d-1}$, \[ \inr{\wh{\mu}_N - \mu, u}\le \frac{C}{\sqrt{N}} \left( \sigma(u)\sqrt{\log(1/\delta)} + \left(\E\|X-\EXP X\|_2^2\right)^{1/2} \right)~, \] where $\sigma^2(u) = \var(\inr{X,u})$ 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.
期刊介绍:
The Journal of the European Mathematical Society (JEMS) is the official journal of the EMS.
The Society, founded in 1990, works at promoting joint scientific efforts between the many different structures that characterize European mathematics. JEMS will publish research articles in all active areas of pure and applied mathematics. These will be selected by a distinguished, international board of editors for their outstanding quality and interest, according to the highest international standards.
Occasionally, substantial survey papers on topics of exceptional interest will also be published. Starting in 1999, the Journal was published by Springer-Verlag until the end of 2003. Since 2004 it is published by the EMS Publishing House. The first Editor-in-Chief of the Journal was J. Jost, succeeded by H. Brezis in 2004.
The Journal of the European Mathematical Society is covered in:
Mathematical Reviews (MR), Current Mathematical Publications (CMP), MathSciNet, Zentralblatt für Mathematik, Zentralblatt MATH Database, Science Citation Index (SCI), Science Citation Index Expanded (SCIE), CompuMath Citation Index (CMCI), Current Contents/Physical, Chemical & Earth Sciences (CC/PC&ES), ISI Alerting Services, Journal Citation Reports/Science Edition, Web of Science.