Current browse context:
math.ST
Change to browse by:
References & Citations
Mathematics > Statistics Theory
Title: Robust Covariance and Scatter Matrix Estimation under Huber's Contamination Model
(Submitted on 1 Jun 2015 (v1), last revised 12 Jun 2017 (this version, v4))
Abstract: Covariance matrix estimation is one of the most important problems in statistics. To accommodate the complexity of modern datasets, it is desired to have estimation procedures that not only can incorporate the structural assumptions of covariance matrices, but are also robust to outliers from arbitrary sources. In this paper, we define a new concept called matrix depth and then propose a robust covariance matrix estimator by maximizing the empirical depth function. The proposed estimator is shown to achieve minimax optimal rate under Huber's $\epsilon$-contamination model for estimating covariance/scatter matrices with various structures including bandedness and sparsity.
Submission history
From: Chao Gao [view email][v1] Mon, 1 Jun 2015 22:08:52 GMT (2788kb,D)
[v2] Fri, 3 Jul 2015 16:06:57 GMT (2789kb,D)
[v3] Thu, 30 Jul 2015 19:38:24 GMT (2781kb)
[v4] Mon, 12 Jun 2017 15:58:11 GMT (90kb,D)
Link back to: arXiv, form interface, contact.