Description: An ADMM based algorithm is developed for estimating a sparse low-rank matrix from its noisy observation. Nonconvex penalties are used with restricted non-convexity so that the total cost function is strictly convex.

The demo.m file contains comparison and a test case.

Please cite as: Improved Sparse and Low-Rank Matrix Estimation. A. Parekh and I. W. Selesnick. Signal Processing, Oct., 2017.
https://doi.org/10.1016/j.sigpro.2017.04.011

Contact: Ankit Parekh (ankit.parekh@nyu.edu)