Cauchy robust principal components analysis with applications to high dimensional datasets

Andrew Wood; Aisha Fayomi; Yannis Pantazis; Michail Tsagris

Cauchy robust principal components analysis with applications to high dimensional datasets

Andrew Wood, Aisha Fayomi, Yannis Pantazis, Michail Tsagris

Research output: Contribution to journal › Article › peer-review

Abstract

In this paper, we propose a modified formulation of the principal components analysis, based on the use of a multivariate Cauchy likelihood instead of the Gaussian likelihood, which has the effect of robustifying the principal components. We present an algorithm to compute these robustified principal components. We additionally derive the relevant influence function of the first component and examine its theoretical properties.

Original language	English
Article number	26
Number of pages	14
Journal	Statistics and Computing
Volume	34
Publication status	Published - 2 Nov 2023

Access to Document

https://link.springer.com/article/10.1007/s11222-023-10328-x

Cite this

@article{7b05d03ea3804a65b5658c5007259e71,

title = "Cauchy robust principal components analysis with applications to high dimensional datasets",

abstract = "In this paper, we propose a modified formulation of the principal components analysis, based on the use of a multivariate Cauchy likelihood instead of the Gaussian likelihood, which has the effect of robustifying the principal components. We present an algorithm to compute these robustified principal components. We additionally derive the relevant influence function of the first component and examine its theoretical properties.",

author = "Andrew Wood and Aisha Fayomi and Yannis Pantazis and Michail Tsagris",

note = "{\textcopyright}TheAuthor(s) 2023",

year = "2023",

month = nov,

day = "2",

language = "English",

volume = "34",

journal = "Statistics and Computing",

issn = "1573-1375",

publisher = "Springer Netherlands",

}

TY - JOUR

T1 - Cauchy robust principal components analysis with applications to high dimensional datasets

AU - Wood, Andrew

AU - Fayomi, Aisha

AU - Pantazis, Yannis

AU - Tsagris, Michail

PY - 2023/11/2

Y1 - 2023/11/2

N2 - In this paper, we propose a modified formulation of the principal components analysis, based on the use of a multivariate Cauchy likelihood instead of the Gaussian likelihood, which has the effect of robustifying the principal components. We present an algorithm to compute these robustified principal components. We additionally derive the relevant influence function of the first component and examine its theoretical properties.

AB - In this paper, we propose a modified formulation of the principal components analysis, based on the use of a multivariate Cauchy likelihood instead of the Gaussian likelihood, which has the effect of robustifying the principal components. We present an algorithm to compute these robustified principal components. We additionally derive the relevant influence function of the first component and examine its theoretical properties.

M3 - Article

SN - 1573-1375

VL - 34

JO - Statistics and Computing

JF - Statistics and Computing

M1 - 26

ER -

Cauchy robust principal components analysis with applications to high dimensional datasets

Abstract

Access to Document

Fingerprint

Cite this