Adversarial Robust Low Rank Matrix Estimation: Compressed Sensing and Matrix Completion

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Sasai, Takeyuki, Fujisawa, Hironori
Formato: Preprint
Publicado: 2020
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866929355444191232
author Sasai, Takeyuki
Fujisawa, Hironori
author_facet Sasai, Takeyuki
Fujisawa, Hironori
contents We consider robust low rank matrix estimation as a trace regression when outputs are contaminated by adversaries. The adversaries are allowed to add arbitrary values to arbitrary outputs. Such values can depend on any samples. We deal with matrix compressed sensing, including lasso as a partial problem, and matrix completion, and then we obtain sharp estimation error bounds. To obtain the error bounds for different models such as matrix compressed sensing and matrix completion, we propose a simple unified approach based on a combination of the Huber loss function and the nuclear norm penalization, which is a different approach from the conventional ones. Some error bounds obtained in the present paper are sharper than the past ones.
format Preprint
id arxiv_https___arxiv_org_abs_2010_13018
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Adversarial Robust Low Rank Matrix Estimation: Compressed Sensing and Matrix Completion
Sasai, Takeyuki
Fujisawa, Hironori
Machine Learning
Statistics Theory
62G35, 62G05
We consider robust low rank matrix estimation as a trace regression when outputs are contaminated by adversaries. The adversaries are allowed to add arbitrary values to arbitrary outputs. Such values can depend on any samples. We deal with matrix compressed sensing, including lasso as a partial problem, and matrix completion, and then we obtain sharp estimation error bounds. To obtain the error bounds for different models such as matrix compressed sensing and matrix completion, we propose a simple unified approach based on a combination of the Huber loss function and the nuclear norm penalization, which is a different approach from the conventional ones. Some error bounds obtained in the present paper are sharper than the past ones.
title Adversarial Robust Low Rank Matrix Estimation: Compressed Sensing and Matrix Completion
topic Machine Learning
Statistics Theory
62G35, 62G05
url https://arxiv.org/abs/2010.13018