A Pairwise Pseudo-likelihood Approach for Matrix Completion with Informative Missingness

The goal of matrix completion is to recover a target matrix from its noisy and incomplete measurements. It is a modern high-dimensional missing data problem.