We propose Ky Fan 2-k-norm-based models for the nonconvex low-rank matrix
recovery problem. A general difference of convex algorithm (DCA) is developed
to solve these models. Numerical results show that the proposed models achieve
high recoverability rates.