Abstract:Robust matrix completion aims to recover a low-rank matrix from a subset of noisy entries perturbed by complex noises, where traditional methods for matrix completion may perform poorly due to utilizing $l_2$ error norm in optimization. In this paper, we propose a novel and fast robust matrix completion method based on maximum correntropy criterion (MCC). The correntropy based error measure is utilized instead of using $l_2$-based error norm to improve the robustness to noises. Using the half-quadratic optimization technique, the correntropy based optimization can be transformed to a weighted matrix factorization problem. Then, two efficient algorithms are derived, including alternating minimization based algorithm and alternating gradient descend based algorithm. The proposed algorithms do not need to calculate singular value decomposition (SVD) at each iteration. Further, the adaptive kernel selection strategy is proposed to accelerate the convergence speed as well as improve the performance. Comparison with existing robust matrix completion algorithms is provided by simulations, showing that the new methods can achieve better performance than existing state-of-the-art algorithms.

An improved Riemannian conjugate gradient method and its application to robust matrix completion

The convergence of conjugate gradient method with nonmonotone line search

Convergence Analysis of Projected Gradient Descent for Schatten-p Nonconvex Matrix Recovery

Two efficient nonlinear conjugate gradient methods for Riemannian manifolds

Convergence Analysis of Gradient Algorithms on Riemannian Manifolds Without Curvature Constraints and Application to Riemannian Mass

A Modified Dai–Liao Conjugate Gradient Method Based on a Scalar Matrix Approximation of Hessian and Its Application

A Nearly Unbiased Matrix Completion Approach.

Gauss–Newton Method for Convex Composite Optimizations on Riemannian Manifolds

A New Retraction For Accelerating The Riemannian Three-Factor Low-Rank Matrix Completion Algorithm

Global Convergence of Hager-Zhang type Riemannian Conjugate Gradient Method

Robust Matrix Completion via Maximum Correntropy Criterion and Half Quadratic Optimization

Barzilai and Borwein conjugate gradient method equipped with a non-monotone line search technique and its application on non-negative matrix factorization

Riemannian Levenberg-Marquardt Method with Global and Local Convergence Properties

Riemannian preconditioned algorithms for tensor completion via tensor ring decomposition

A sufficient descent three-term conjugate gradient method via symmetric rank-one update for large-scale optimization

A Riemannian rank-adaptive method for low-rank matrix completion

Tensor Completion with Side Information: A Riemannian Manifold Approach

An improved Dai‐Liao‐style hybrid conjugate gradient‐based method for solving unconstrained nonconvex optimization and extension to constrained nonlinear monotone equations

On the extension of the Hager–Zhang conjugate gradient method for vector optimization

Convergence of the Descent Nonlinear Conjugate Gradient Methods

New vector transport operators extending a Riemannian CG algorithm to generalized Stiefel manifold with low-rank applications