Saved in:
Bibliographic Details
Main Authors: Zhang, Lijun, Yang, Tianbao, Jin, Rong, Zhou, Zhi-Hua
Format: Preprint
Published: 2015
Subjects:
Online Access:https://arxiv.org/abs/1504.06817
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909212367388672
author Zhang, Lijun
Yang, Tianbao
Jin, Rong
Zhou, Zhi-Hua
author_facet Zhang, Lijun
Yang, Tianbao
Jin, Rong
Zhou, Zhi-Hua
contents In this paper, we develop a relative error bound for nuclear norm regularized matrix completion, with the focus on the completion of full-rank matrices. Under the assumption that the top eigenspaces of the target matrix are incoherent, we derive a relative upper bound for recovering the best low-rank approximation of the unknown matrix. Although multiple works have been devoted to analyzing the recovery error of full-rank matrix completion, their error bounds are usually additive, making it impossible to obtain the perfect recovery case and more generally difficult to leverage the skewed distribution of eigenvalues. Our analysis is built upon the optimality condition of the regularized formulation and existing guarantees for low-rank matrix completion. To the best of our knowledge, this is the first relative bound that has been proved for the regularized formulation of matrix completion.
format Preprint
id arxiv_https___arxiv_org_abs_1504_06817
institution arXiv
publishDate 2015
record_format arxiv
spellingShingle Relative Error Bound Analysis for Nuclear Norm Regularized Matrix Completion
Zhang, Lijun
Yang, Tianbao
Jin, Rong
Zhou, Zhi-Hua
Machine Learning
In this paper, we develop a relative error bound for nuclear norm regularized matrix completion, with the focus on the completion of full-rank matrices. Under the assumption that the top eigenspaces of the target matrix are incoherent, we derive a relative upper bound for recovering the best low-rank approximation of the unknown matrix. Although multiple works have been devoted to analyzing the recovery error of full-rank matrix completion, their error bounds are usually additive, making it impossible to obtain the perfect recovery case and more generally difficult to leverage the skewed distribution of eigenvalues. Our analysis is built upon the optimality condition of the regularized formulation and existing guarantees for low-rank matrix completion. To the best of our knowledge, this is the first relative bound that has been proved for the regularized formulation of matrix completion.
title Relative Error Bound Analysis for Nuclear Norm Regularized Matrix Completion
topic Machine Learning
url https://arxiv.org/abs/1504.06817