Abstract:High-dimensional matrix regression has been studied in various aspects, such as statistical properties, computational efficiency and application to specific instances including multivariate regression, system identification and matrix compressed sensing. Current studies mainly consider the idealized case that the covariate matrix is obtained without noise, while the more realistic scenario that the covariates may always be corrupted with noise or missing data has received little attention. We consider the general errors-in-variables matrix regression model and proposed a unified framework for low-rank estimation based on nonconvex spectral regularization. Then in the statistical aspect, recovery bounds for any stationary points are provided to achieve statistical consistency. In the computational aspect, the proximal gradient method is applied to solve the nonconvex optimization problem and is proved to converge in polynomial time. Consequences for specific matrix compressed sensing models with additive noise and missing data are obtained via verifying corresponding regularity conditions. Finally, the performance of the proposed nonconvex estimation method is illustrated by numerical experiments.

What problem does this paper attempt to address?

This paper attempts to solve the low - rank matrix estimation problem in high - dimensional matrix regression when independent variables (covariates) are affected by noise or missing data. Specifically, most of the existing studies assume that the independent variables are noise - free, while in reality, the independent variables are often contaminated by noise or have missing values. In this case, traditional low - rank matrix estimation methods may fail. Therefore, this paper proposes a unified framework based on non - convex spectral regularization to deal with this problem and provides theoretical guarantees for statistical consistency and algorithm convergence. ### Main contributions: 1. **Propose a new non - convex spectral regularization method**: This method can effectively perform low - rank matrix estimation when the independent variables are affected by noise. 2. **Establish statistical recovery bounds**: Provide statistical consistency guarantees for any stable point of the non - convex optimization problem. 3. **Prove the linear convergence of the algorithm**: Modify the proximal gradient method and prove its effectiveness in linearly converging to the global solution. 4. **Verify the conditions under specific models**: Through probability analysis, verify that the proposed conditions hold in specific error - in - variables models. ### Technical details: - **Non - convex spectral regularization**: Regularize the singular values of the matrix by non - convex functions to reduce the bias caused by measurement errors. - **Restricted strong convexity and restricted strong smoothness**: Propose the restricted strong convexity and restricted strong smoothness conditions applicable to error - in - variables matrix regression and prove that these conditions hold in high - dimensional cases. - **Proximal gradient method**: Modify the proximal gradient method so that it can effectively solve non - convex optimization problems and prove its linear convergence. ### Application scenarios: - **Multi - response model**: Deal with the situation of multiple response variables. - **Matrix compressed sensing**: Widely used in signal processing and image reconstruction. - **Matrix completion**: Deal with the situation of partially observed data. ### Conclusion: The method proposed in this paper not only has good statistical consistency and algorithm convergence theoretically, but also shows excellent performance in numerical experiments. This provides a new and effective tool for dealing with the error - in - variables problem in high - dimensional matrix regression.

Low-rank matrix estimation via nonconvex spectral regularized methods in errors-in-variables matrix regression

Low-rank matrix recovery via nonconvex optimization methods with application to errors-in-variables matrix regression

Low-rank matrix estimation via nonconvex optimization methods in multi-response errors-in-variables regression

Low-rank matrix estimation in multi-response regression with measurement errors: Statistical and computational guarantees.

Convergence of Projected Subgradient Method with Sparse or Low-Rank Constraints

Robust Regularized Low-Rank Matrix Models for Regression and Classification

A Unified Computational and Statistical Framework for Nonconvex Low-rank Matrix Estimation

Regularized Matrix Regression

Low-rank matrix recovery with non-quadratic loss: projected gradient method and regularity projection oracle

Matrix recovery from nonconvex regularized least absolute deviations

A framework of regularized low-rank matrix models for regression and classification

Fast Low-Rank Matrix Learning with Nonconvex Regularization

Estimation of Sparse Covariance Matrix Via Non-Convex Regularization

Low-Rank Matrix Completion Based on Maximum Likelihood Estimation.

Simultaneously Sparse and Low-Rank Matrix Reconstruction Via Nonconvex and Nonseparable Regularization

Low rank matrix recovery using unconstrained smoothed lq minimization

Nonconvex Robust Low-rank Matrix Recovery

Improved Nonconvex Optimization Model for Low-Rank Matrix Recovery

Recovery of corrupted low-rank matrices via half-quadratic based nonconvex minimization

A Unified Variance Reduction-Based Framework for Nonconvex Low-Rank Matrix Recovery

Elastic-Net Regularization For Low-Rank Matrix Recovery