Unbiased Risk Estimates for Singular Value Thresholding and Spectral Estimators

Type: Article

Publication Date: 2013-06-21

Citations: 222

DOI: https://doi.org/10.1109/tsp.2013.2270464

Abstract

In an increasing number of applications, it is of interest to recover an approximately low-rank data matrix from noisy observations. This paper develops an unbiased risk estimate---holding in a Gaussian model---for any spectral estimator obeying some mild regularity assumptions. In particular, we give an unbiased risk estimate formula for singular value thresholding (SVT), a popular estimation strategy which applies a soft-thresholding rule to the singular values of the noisy observations. Among other things, our formulas offer a principled and automated way of selecting regularization parameters in a variety of problems. In particular, we demonstrate the utility of the unbiased risk estimation for SVT-based denoising of real clinical cardiac MRI series data. We also give new results concerning the differentiability of certain matrix-valued functions.

Locations

  • IEEE Transactions on Signal Processing - View - PDF
  • arXiv (Cornell University) - View - PDF
  • DataCite API - View

Similar Works

Action Title Year Authors
+ PDF Chat OptShrink: An Algorithm for Improved Low-Rank Signal Matrix Denoising by Optimal, Data-Driven Singular Value Shrinkage 2014 Raj Rao Nadakuditi
+ Adaptive Shrinkage of singular values 2013 Julie Josse
Sylvain Sardy
+ Adaptive Shrinkage of singular values 2013 Julie Josse
Sylvain Sardy
+ Data-Driven optimal shrinkage of singular values under high-dimensional noise with separable covariance structure with application 2022 Pei-Chun Su
Hau‐Tieng Wu
+ PDF Chat Signal denoising using the minimum-probability-of-error criterion 2020 Jishnu Sadasivan
Subhadip Mukherjee
Chandra Sekhar Seelamantula
+ Generalized SURE for optimal shrinkage of singular values in low-rank matrix denoising 2016 Jérémie Bigot
Charles Deledalle
Delphine FĂ©ral
+ Selecting thresholding and shrinking parameters with generalized SURE for low rank matrix estimation 2013 Julie Josse
Sylvain Sardy
+ The Optimal Hard Threshold for Singular Values is 4/sqrt(3) 2013 Matan Gavish
David L. Donoho
+ The Optimal Hard Threshold for Singular Values is 4/sqrt(3) 2013 Matan Gavish
David L. Donoho
+ Minimax risk of matrix denoising by singular value thresholding 2014 David L. Donoho
Matan Gavish
+ Risk Estimation Without Using Stein's Lemma -- Application to Image Denoising 2014 Sagar Gubbi
Chandra Sekhar Seelamantula
+ STEIN UNBIASED RISK ESTIMATE AS A MODEL SELECTION ESTIMATOR 2020 Nihad Nouri
Fatiha Mezoued
+ Optimal singular value shrinkage for operator norm loss 2020 William Leeb
+ Stability and Risk Bounds of Iterative Hard Thresholding 2022 Xiao‐Tong Yuan
Ping Li
+ PDF Chat Stochastic Iterative Hard Thresholding for Low-Tucker-Rank Tensor Recovery 2020 Rachel Grotheer
Shuang Li
Anna Ma
Deanna Needell
Jing Qin
+ ADA-PT: An Adaptive Parameter Tuning Strategy Based on the Weighted Stein Unbiased Risk Estimator 2018 Rita Ammanouil
André Ferrari
CĂ©dric Richard
+ Stochastic Iterative Hard Thresholding for Low-Tucker-Rank Tensor Recovery 2019 Rachel Grotheer
Shuang Li
Anna Ma
Deanna Needell
Jing Qin
+ PDF Chat Stein Unbiased GrAdient estimator of the Risk (SUGAR) for Multiple Parameter Selection 2014 Charles‐Alban Deledalle
Samuel Vaiter
Jalal Fadili
Gabriel Peyré
+ An iterative hard thresholding estimator for low rank matrix recovery with explicit limiting distribution 2015 Alexandra Carpentier
Arlene K. H. Kim
+ An iterative hard thresholding estimator for low rank matrix recovery with explicit limiting distribution 2015 Alexandra Carpentier
Arlene K. H. Kim

Works That Cite This (118)

Action Title Year Authors
+ PDF Chat Second-order Stein: SURE for SURE and other applications in high-dimensional inference 2021 Pierre Bellec
Cun‐Hui Zhang
+ PDF Chat Spectral operators of matrices 2017 Chao Ding
Defeng Sun
Jie Sun
Kim-Chuan Toh
+ Multinomial Multiple Correspondence Analysis 2016 Patrick J. F. Groenen
Julie Josse
+ PDF Chat The utility of multivariate outlier detection techniques for data quality evaluation in large studies: an application within the ONDRI project 2019 Kelly M. Sunderland
Derek Beaton
Julia Fraser
Donna Kwan
Paula McLaughlin
Manuel Montero‐Odasso
Alicia Peltsch
Frederico Pieruccini‐Faria
Demetrios J. Sahlas
Richard H. Swartz
+ Some considerations on the vibrational environment of the DSC-DCMIX1 experiment onboard ISS 2016 R. Jurado
Jna. GavaldĂ 
Michael Simon
Jordi PallarĂšs
Ana LaverĂłn-Simavilla
Xavier RuĂ­z
Valentina Shevtsova
+ Optimal High-Order Tensor SVD via Tensor-Train Orthogonal Iteration 2022 Yuchen Zhou
Anru R. Zhang
Lili Zheng
Yazhen Wang
+ PDF Chat Imputation and low-rank estimation with Missing Not At Random data 2020 Aude Sportisse
Claire Boyer
Julie Josse
+ A rank-two relaxed parallel splitting version of the augmented Lagrangian method with step size in (0,2) for separable convex programming 2023 Bingsheng He
Feng Ma
Shengjie Xu
Xiaoming Yuan
+ Generalized Factor Model for Ultra-High Dimensional Correlated Variables with Mixed Types 2021 Wei Liu
Huazhen Lin
Shurong Zheng
Jin Liu
+ PDF Chat SURE Based Truncated Tensor Nuclear Norm Regularization for Low Rank Tensor Completion 2020 Gordon Morison