Signal processing algorithms

Nonconvex Nonsmooth Low-Rank Minimization via Iteratively Reweighted Nuclear Norm

We propose to use a family of nonconvex surrogates of L 0 -norm on the singular values of a matrix to approximate the rank function.