Large Dimensional Analysis of Robust M-Estimators of Covariance With Outliers

DOI10.1109/TSP.2015.2460225MaRDI QIDQ4580893zbMATH OpenOpenAlexFDO

Authors David Morales-Jimenez, R. Couillet, Matthew McKay

Publication date 22 August 2018

Published in IEEE Transactions on Signal Processing (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1503.01245

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Estimation in multivariate analysis (62H12) Signal theory (characterization, reconstruction, filtering, etc.) (94A12) Robustness and adaptive procedures (parametric inference) (62F35)

Abstract: A large dimensional characterization of robust M-estimators of covariance (or scatter) is provided under the assumption that the dataset comprises independent (essentially Gaussian) legitimate samples as well as arbitrary deterministic samples, referred to as outliers. Building upon recent random matrix advances in the area of robust statistics, we specifically show that the so-called Maronna M-estimator of scatter asymptotically behaves similar to well-known random matrices when the population and sample sizes grow together to infinity. The introduction of outliers leads the robust estimator to behave asymptotically as the weighted sum of the sample outer products, with a constant weight for all legitimate samples and different weights for the outliers. A fine analysis of this structure reveals importantly that the propensity of the M-estimator to attenuate (or enhance) the impact of outliers is mostly dictated by the alignment of the outliers with the inverse population covariance matrix of the legitimate samples. Thus, robust M-estimators can bring substantial benefits over more simplistic estimators such as the per-sample normalized version of the sample covariance matrix, which is not capable of differentiating the outlying samples. The analysis shows that, within the class of Maronna's estimators of scatter, the Huber estimator is most favorable for rejecting outliers. On the contrary, estimators more similar to Tyler's scale invariant estimator (often preferred in the literature) run the risk of inadvertently enhancing some outliers.

Cited in

(6)

• Regression analysis: likelihood, error and entropy
• Efficient computation of limit spectra of sample covariance matrices
• Robust spiked random matrices and a robust G-MUSIC estimator
• Tyler's and Maronna's M-estimators: non-asymptotic concentration results
• Marčenko-Pastur law for Tyler's M-estimator
• Robust sparse covariance estimation by thresholding Tyler's M-estimator

This page was built for publication: Large Dimensional Analysis of Robust M-Estimators of Covariance With Outliers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4580893)