Abstract:A popular approach for estimating an unknown signal from noisy, linear measurements is via solving a so called \emph{regularized M-estimator}, which minimizes a weighted combination of a convex loss function and of a convex (typically, non-smooth) regularizer. We accurately predict the squared error performance of such estimators in the high-dimensional proportional regime. The random measurement matrix is assumed to have entries iid Gaussian, only minimal and rather mild regularity conditions are imposed on the loss function, the regularizer, and on the noise and signal distributions. We show that the error converges in probability to a nontrivial limit that is given as the solution to a minimax convex-concave optimization problem on four scalar optimization variables. We identify a new summary parameter, termed the Expected Moreau envelope to play a central role in the error characterization. The \emph{precise} nature of the results permits an accurate performance comparison between different instances of regularized M-estimators and allows to optimally tune the involved parameters (e.g. regularizer parameter, number of measurements). The key ingredient of our proof is the \emph{Convex Gaussian Min-max Theorem} (CGMT) which is a tight and strengthened version of a classical Gaussian comparison inequality that was proved by Gordon in 1988.

An Accurate Perturbation Analysis Algorithm For Music With Toeplitz Covariance Matrix

Theoretical Comparison For Biases Of Music-Like Doa Estimators

Toeplitz covariance estimation with applications to MUSIC

The Optimum Weight of Angle-Dependent Weighted MUSIC and Its Approximations

Design of Two MUSIC-like Estimators Based on Bias Minimization

Iterative Multiple Signal Classification Algorithm with Small Sample Size

Asymptotic Performance Analysis of the MUSIC Algorithm for Direction-of-Arrival Estimation

The robust covariation-based MUSIC (ROC-MUSIC) algorithm for bearing estimation in impulsive noise environments

A performance analysis of subspace-based methods in the presence of model errors. I. The MUSIC algorithm

Appearance of inaccurate results in the MUSIC algorithm with inappropriate wavenumber

A Subspace Estimator for Fixed Rank Perturbations of Large Random Matrices

Analysis of the asymptotic relative efficiency of the MUSIC algorithm

Quantum algorithm for MUSIC-based DOA estimation in hybrid MIMO systems

A Numerically Robust Method for DOA Estimation

Linear Time-Frequency Music Algorithm

An improved root-MUSIC algorithm and MSE analysis

Stable subspace dimension reduced MUSIC for blade tip timing

Fast Computation of Real Polynomial Coefficients for Spectral Capon/MUSIC Rooting Algorithm

Precise Error Analysis of Regularized M-estimators in High-dimensions

Music statistics: uncertain logistic regression models with applications in analyzing music

A stochastic perturbation analysis of the QR decomposition and its applications