Low-rank matrix recovery with Ky Fan 2-k-norm

Abstract Low-rank matrix recovery problem is difficult due to its non-convex properties and it usually solved using convex relaxation approaches. In this paper, we formulate the low-rank exactly novel Ky Fan 2- k -norm-based models. A general difference of functions algorithm (DCA) developed solve these proximal point (PPA) framework proposed sub-problems within DCA, which allows us handle large instances. Numerical results show that models achieve high recoverability rates as compared truncated nuclear norm method alternating bilinear optimization approach. The also demonstrate DCA with PPA efficient in handling larger