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.

Asymptotic properties for M-estimators in linear models with dependent random errors

Asymptotics of M-estimator in Multivariate Linear Regression Models for a Class of Random Errors

ROBUST ESTIMATION IN PARAMETRIC TIME SERIES MODELS UNDER LONG‐ AND SHORT‐RANGE‐DEPENDENT STRUCTURES

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

Nonparametric M-estimation for Functional Stationary Ergodic Data

Asymptotic Property of M Estimator in Classical Linear Models under Dependent Random Errors

Asymptotic Behavior for S-estimators in Random Design Linear Model with Long-Range-dependent Errors

Asymptotic Properties of Nonparametric M-Estimation for Mixing Functional Data

The Strong Consistency of M Estimator in Linear Models Based on Widely Orthant Dependent Errors

M-estimation for linear models with exchangeable errors

Asymptotic properties of <i>M</i> estimators in classical linear models with <i>f</i>-mixing random errors

The Strong Consistency of M-estimates in Linear Models with Extended Negatively Dependent Errors

Strong Approximations of Martingale Vectors and Their Applications in Markov-Chain Adaptive Designs

Asymptotic Properties of the Wavelet Estimator in Non Parametric Regression Model with Martingale Difference Errors

An Exponential Inequality and Its Application to M Estimators in Multiple Linear Models

Asymptotic Normality in Partial Linear Models Based on Dependent Errors

The Asymptotic Properties for the Estimators in a Semiparametric Regression Model Based on M-Asymptotic Negatively Associated Errors

Asymptotic Properties for an M-estimator of the Regression Function with Truncation and Dependent Data

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

Asymptotics for A Class of Dependent Random Variables

Limiting behavior of recursive M-estimators in multivariate linear regression models and their asymptotic efficiencies