We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Covariance Estimation: Optimal Dimension-free Guarantees for Adversarial Corruption and Heavy Tails

Abstract: We provide an estimator of the covariance matrix that achieves the optimal rate of convergence (up to constant factors) in the operator norm under two standard notions of data contamination: We allow the adversary to corrupt an $\eta$-fraction of the sample arbitrarily, while the distribution of the remaining data points only satisfies that the $L_{p}$-marginal moment with some $p \ge 4$ is equivalent to the corresponding $L_2$-marginal moment. Despite requiring the existence of only a few moments, our estimator achieves the same tail estimates as if the underlying distribution were Gaussian. As a part of our analysis, we prove a dimension-free Bai-Yin type theorem in the regime $p > 4$.
Comments: 32 pages, minor changes, added a reference to a forthcoming parallel work of Z. Fernandez-Rico and R. I. Oliveira
Subjects: Statistics Theory (math.ST); Data Structures and Algorithms (cs.DS); Probability (math.PR)
Cite as: arXiv:2205.08494 [math.ST]
  (or arXiv:2205.08494v2 [math.ST] for this version)

Submission history

From: Nikita Zhivotovskiy [view email]
[v1] Tue, 17 May 2022 17:08:45 GMT (31kb)
[v2] Wed, 22 Jun 2022 18:09:52 GMT (33kb)

Link back to: arXiv, form interface, contact.