Alternating Iteratively Reweighted Least Squares Minimization for Low-Rank Matrix Factorization

Nowadays, the availability of large-scale data in disparate application domains urges the deployment of sophisticated tools for extracting valuable knowledge out of this huge bulk of information. In that vein, low-rank representations (LRRs), which seek low-dimensional embeddings of data have naturally appeared. In an effort to reduce computational complexity and improve estimation performance, LRR has been viewed via a matrix factorization (MF) perspective. Recently, low-rank MF (LRMF) approaches have been proposed for tackling the inherent weakness of MF, i.e., the unawareness of the dimension of the low-dimensional space where data reside. Herein, inspired by the merits of iterative reweighted schemes for sparse recovery and rank minimization, we come up with a generic low-rank promoting regularization function. Then, focusing on a specific instance of it, we propose a regularizer that imposes column-sparsity jointly on the two matrix factors that result from MF, thus promoting low-rankness on the optimization problem. The low-rank promoting properties of the resulting regularization term are brought to light by mathematically showing that it is actually a tight upper bound of a specific version of the weighted nuclear norm. The problems of denoising and matrix completion are redefined according to the new LRMF formulation and solved via efficient alternating iteratively reweighted least squares type algorithms. Theoretical analysis of the algorithms regarding the convergence and the rates of convergence to stationary points is provided. The effectiveness of the proposed algorithms is verified in diverse simulated and real data experiments.