Abstract:Heavy-tailed errors impair the accuracy of the least squares estimate, which can be spoiled by a single grossly outlying observation. As argued in the seminal work of Peter Huber in 1973 [Ann. Statist.1 (1973) 799-821], robust alternatives to the method of least squares are sorely needed. To achieve robustness against heavy-tailed sampling distributions, we revisit the Huber estimator from a new perspective by letting the tuning parameter involved diverge with the sample size. In this paper, we develop nonasymptotic concentration results for such an adaptive Huber estimator, namely, the Huber estimator with the tuning parameter adapted to sample size, dimension, and the variance of the noise. Specifically, we obtain a sub-Gaussian-type deviation inequality and a nonasymptotic Bahadur representation when noise variables only have finite second moments. The nonasymptotic results further yield two conventional normal approximation results that are of independent interest, the Berry-Esseen inequality and Cramér-type moderate deviation. As an important application to large-scale simultaneous inference, we apply these robust normal approximation results to analyze a dependence-adjusted multiple testing procedure for moderately heavy-tailed data. It is shown that the robust dependence-adjusted procedure asymptotically controls the overall false discovery proportion at the nominal level under mild moment conditions. Thorough numerical results on both simulated and real datasets are also provided to back up our theory.

An Edgeworth expansion for finite population L-statistics

Edgeworth approximations for distributions of symmetric statistics

Edgeworth Expansion by Stein's Method

Edgeworth expansions for a sample sum from a finite set of independent random variables

Asymptotic expansion for nonparametric M-estimator in a nonlinear regression model with long-memory errors

Precise large deviations for randomly weighted sums of negatively dependent random variables with consistently varying tails

Fractional Edgeworth expansions for one-dimensional heavy-tailed random variables and applications

Tail Asymptotic Expansions for L-Statistics

Edgeworth corrections for spot volatility estimator

Edgeworth expansion for the pre-averaging estimator

Edgeworth expansion for the coefficients of random walks on the general linear group

5th-Order Multivariate Edgeworth Expansions for Parametric Estimates

Asymptotic Expansions in the CLT in Free Probability

A NEW PERSPECTIVE ON ROBUST M-ESTIMATION: FINITE SAMPLE THEORY AND APPLICATIONS TO DEPENDENCE-ADJUSTED MULTIPLE TESTING

Asymptotic Approximation of Inverse Moments of Nonnegative Random Variables

Comparing Edgeworth Expansion and Saddlepoint Approximation in Assessing the Asymptotic Normality Behavior of A Non-Parametric Estimator for Finite Population Total

Asymptotic expansions relating to the distribution of the length of longest increasing subsequences

Tail bounds for empirically standardized sums

Fixed-smoothing asymptotics for time series

Moderate Deviations for Finite Population Student's Statistic

An Analysis Of Estimation Of Distribution Algorithms With Finite Population Models