Frame-constrained total variation regularization for white noise regression

DOI10.1214/20-AOS2001zbMATH Open1475.62145arXiv1807.02038MaRDI QIDQ820796FDOQ820796

Authors: Miguel Del Álamo, Housen Li, Axel Munk 

Publication date: 28 September 2021

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: Despite the popularity and practical success of total variation (TV) regularization for function estimation, surprisingly little is known about its theoretical performance in a statistical setting. While TV regularization has been known for quite some time to be minimax optimal for denoising one-dimensional signals, for higher dimensions this remains elusive until today. In this paper we consider frame-constrained TV estimators including many well-known (overcomplete) frames in a white noise regression model, and prove their minimax optimality w.r.t. $L^q$-risk ($1leqq<infty$) up to a logarithmic factor in any dimension $dgeq1$. Overcomplete frames are an established tool in mathematical imaging and signal recovery, and their combination with TV regularization has been shown to give excellent results in practice, which our theory now confirms. Our results rely on a novel connection between frame-constraints and certain Besov norms, and on an interpolation inequality to relate them to the risk functional.

Full work available at URL: https://arxiv.org/abs/1807.02038

Recommendations

• Optimal convergence rates for Tikhonov regularization in Besov spaces
• Optimal convergence rates for sparsity promoting wavelet-regularization in Besov spaces
• Deconvolution model with fractional Gaussian noise: a minimax study
• Noise estimation in signal restoration using regularization
• Convergence rates for oversmoothing Banach space regularization

zbMATH Keywords

total variationwaveletsnonparametric regressionminimax estimationinterpolation inequalitiesovercomplete dictionaries

Mathematics Subject Classification ID

Nonparametric regression and quantile regression (62G08) Asymptotic properties of nonparametric inference (62G20) Random fields; image analysis (62M40) Minimax procedures in statistical decision theory (62C20) Numerical methods for wavelets (65T60) White noise theory (60H40)

Cites Work

• Local extremes, runs, strings and multiresolution. (With discussion)
• Nonlinear total variation based noise removal algorithms
• Title not available (Why is that?)
• Title not available (Why is that?)
• Locally adaptive regression splines
• Multiscale testing of qualitative hypotheses
• Adaptive risk bounds in univariate total variation denoising and trend filtering
• Multiscale change-point segmentation: beyond step functions
• Mathematical Foundations of Infinite-Dimensional Statistical Models
• Ten Lectures on Wavelets
• Multiscale Change Point Inference
• Introduction to nonparametric estimation
• Multivariate extensions of isotonic regression and total variation denoising via entire monotonicity and Hardy-Krause variation
• Title not available (Why is that?)
• Asymptotic equivalence of nonparametric regression and white noise
• Extensions of smoothing via taut strings
• On the prediction performance of the Lasso
• Image recovery via total variation minimization and related problems
• An algorithm for total variation minimization and applications
• A first-order primal-dual algorithm for convex problems with applications to imaging
• Discretization-invariant Bayesian inversion and Besov space priors
• Oscillating patterns in image processing and nonlinear evolution equations. The fifteenth Dean Jacqueline B. Lewis memorial lectures
• Title not available (Why is that?)
• Image Decomposition and Restoration Using Total Variation Minimization and theH¹
• Can one use total variation prior for edge-preserving Bayesian inversion?
• Variational methods in imaging
• Minimax estimation via wavelet shrinkage
• Title not available (Why is that?)
• On adaptive minimax density estimation on \(\mathbb R^d\)
• New tight frames of curvelets and optimal representations of objects with piecewise C² singularities
• Asymptotic equivalence for nonparametric regression with multivariate and random design
• Practical Approximate Solutions to Linear Operator Equations When the Data are Noisy
• Title not available (Why is that?)
• Optimal spatial adaptation to inhomogeneous smoothness: An approach based on kernel estimates with variable bandwidth selectors
• Automated regularization parameter selection in multi-scale total variation models for image restoration
• Sparse multidimensional representations using anisotropic dilation and shear operators
• A semismooth Newton method for \(\mathrm{L}^1\) data fitting with automatic choice of regularization parameters and noise calibration
• Nonparametric estimation of scalar diffusions based on low frequency data
• On Difference-Based Variance Estimation in Nonparametric Regression When the Covariate is High Dimensional
• Frame decomposition of decomposition spaces
• Harmonic analysis of the space BV.
• On improved Sobolev embedding theorems
• Image decompositions using bounded variation and generalized homogeneous Besov spaces
• On a Problem of Adaptive Estimation in Gaussian White Noise
• On minimax wavelet estimators
• Shearlet smoothness spaces
• Unconditional bases are optimal bases for data compression and for statistical estimation
• Title not available (Why is that?)
• Characterizations of Besov-Hardy-Sobolev spaces: A unified approach
• New multiscale transforms, minimum total variation synthesis: Applications to edge-preserving image reconstruction
• Statistical multiresolution Dantzig estimation in imaging: fundamental concepts and algorithmic framework
• Statistical multiresolution estimation for variational imaging: with an application in Poisson-biophotonics
• Variance estimation for high-dimensional regression models
• Variational estimators in statistical multiscale analysis
• Variational multiscale nonparametric regression: smooth functions
• Nonparametric estimation over shrinking neighborhoods: superefficiency and adaptation
• Multiscale scanning in inverse problems
• Isotonic regression in multi-dimensional spaces and graphs
• Isotonic regression in general dimensions
• Adaptive estimation over anisotropic functional classes via oracle approach
• Extreme value analysis of empirical frame coefficients and implications for denoising by soft-thresholding
• Discrete Total Variation: New Definition and Minimization
• Multiresolution Parameter Choice Method for Total Variation Regularized Tomography
• Total variation regularization of multi-material topology optimization
• An improvement of Rudin-Osher-Fatemi model

Cited In (4)

• Multiscale change-point segmentation: beyond step functions
• Multivariate trend filtering for lattice data
• Convergence rates for oversmoothing Banach space regularization
• Spline smoothing in regression models and asymptotic efficiency in \(L_ 2\)

This page was built for publication: Frame-constrained total variation regularization for white noise regression

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q820796)