Robust Lasso With Missing and Grossly Corrupted Observations

DOI10.1109/TIT.2012.2232347zbMATH Open1364.94146arXiv1112.0391OpenAlexW2057690238MaRDI QIDQ2989299FDOQ2989299

Authors:

Publication date: 8 June 2017

Published in: IEEE Transactions on Information Theory (Search for Journal in Brave)

Abstract: This paper studies the problem of accurately recovering a sparse vector $$ from highly corrupted linear measurements $$ where $e^{star}$ is a sparse error vector whose nonzero entries may be unbounded and $w$ is a bounded noise. We propose a so-called extended Lasso optimization which takes into consideration sparse prior information of both $$ and $e^{star}$. Our first result shows that the extended Lasso can faithfully recover both the regression as well as the corruption vector. Our analysis relies on the notion of extended restricted eigenvalue for the design matrix $X$. Our second set of results applies to a general class of Gaussian design matrix $X$ with i.i.d rows $operN(0,Sigma)$, for which we can establish a surprising result: the extended Lasso can recover exact signed supports of both $$ and $e^{star}$ from only $Omega(klogplogn)$ observations, even when the fraction of corruption is arbitrarily close to one. Our analysis also shows that this amount of observations required to achieve exact signed support is indeed optimal.

Full work available at URL: https://arxiv.org/abs/1112.0391

Mathematics Subject Classification ID

Statistical aspects of information-theoretic topics (62B10) Signal theory (characterization, reconstruction, filtering, etc.) (94A12)

Cited In (21)

• Joint outlier detection and variable selection using discrete optimization
• Trustworthy regularized huber regression for outlier detection
• A general family of trimmed estimators for robust high-dimensional data analysis
• Robust and tuning-free sparse linear regression via square-root slope
• Title not available (Why is that?)
• Confidence regions and minimax rates in outlier-robust estimation on the probability simplex
• Uniform recovery guarantees for quantized corrupted sensing using structured or generative priors
• A Sharper Computational Tool for Regression
• High-dimensional inference robust to outliers with ℓ1-norm penalization
• Covariate Selection in High-Dimensional Generalized Linear Models With Measurement Error
• Multidimensional linear functional estimation in sparse Gaussian models and robust estimation of the mean
• Outlier detection and robust mixture modeling using nonconvex penalized likelihood
• A User-Friendly Computational Framework for Robust Structured Regression with the L₂ Criterion
• Provable training set debugging for linear regression
• Robust subset selection
• Geometric median and robust estimation in Banach spaces
• Iterative gradient descent for outlier detection
• Linear regression with sparsely permuted data
• Computational and statistical analyses for robust non-convex sparse regularized regression problem
• Robust moderately clipped LASSO for simultaneous outlier detection and variable selection
• Robust regression with compositional covariates

This page was built for publication: Robust Lasso With Missing and Grossly Corrupted Observations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2989299)