Spatial Sign based Principal Component Analysis for High Dimensional Data

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhao, Ping, Wang, Hongfei, Feng, Long
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909676264751104
author Zhao, Ping
Wang, Hongfei
Feng, Long
author_facet Zhao, Ping
Wang, Hongfei
Feng, Long
contents This article focuses on the robust principal component analysis (PCA) of high-dimensional data with elliptical distributions. We investigate the PCA of the sample spatial-sign covariance matrix in both nonsparse and sparse contexts, referring to them as SPCA and SSPCA, respectively. We present both nonasymptotic and asymptotic analyses to quantify the theoretical performance of SPCA and SSPCA. In sparse settings, we demonstrate that SSPCA, implemented through a combinatoric program, achieves the optimal rate of convergence. Our proposed SSPCA method is computationally efficient and exhibits robustness against heavy-tailed distributions compared to existing methods. Simulation studies and real-world data applications further validate the superiority of our approach.
format Preprint
id arxiv_https___arxiv_org_abs_2409_13267
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Spatial Sign based Principal Component Analysis for High Dimensional Data
Zhao, Ping
Wang, Hongfei
Feng, Long
Methodology
This article focuses on the robust principal component analysis (PCA) of high-dimensional data with elliptical distributions. We investigate the PCA of the sample spatial-sign covariance matrix in both nonsparse and sparse contexts, referring to them as SPCA and SSPCA, respectively. We present both nonasymptotic and asymptotic analyses to quantify the theoretical performance of SPCA and SSPCA. In sparse settings, we demonstrate that SSPCA, implemented through a combinatoric program, achieves the optimal rate of convergence. Our proposed SSPCA method is computationally efficient and exhibits robustness against heavy-tailed distributions compared to existing methods. Simulation studies and real-world data applications further validate the superiority of our approach.
title Spatial Sign based Principal Component Analysis for High Dimensional Data
topic Methodology
url https://arxiv.org/abs/2409.13267