点击次数：
论文类型：期刊论文
发表时间：2015-07-04
发表刊物：OPTIMIZATION METHODS & SOFTWARE
收录刊物：SCIE、EI
卷号：30
期号：4
页面范围：682-705
ISSN号：1055-6788
关键字：low-rank problem; l(2)(2)-l(p)(p) minimization; majorization method; lower bound analysis; smoothing method
摘要：In this paper, we consider the l(2)(2)-l(p)(p) (with p is an element of(0, 1)) matrix minimization for recovering the low-rank matrices. A smoothing approach for solving this non-smooth, non-Lipschitz and non-convex l(2)(2)-l(p)(p) optimization problem is developed, in which the smoothing parameter is treated as a decision variable and a majorization method is adopted to solve the smoothing problem. The convergence theorem shows that any accumulation point of the sequence generated by the proposed approach satisfies the first-order necessary optimality condition of the l(2)(2)-l(p)(p) problem. As an application, we use the proposed smoothing majorization method to solve the famous matrix completion problems. Numerical results indicate that our algorithm can solve the test problems efficiently.