所属单位：

理学院

发表刊物：

ICIC Express Letters, Part B: Applications

关键字：

中文关键字：矩阵补全；广义低秩矩阵逼近；交替方向法；三分解；压缩率，英文关键字：Matrix completion; generalized low rank approximat

摘要：

Matrix completion tasks are usually achieved by the aid of low rank structure of the underlying data. For the data collection of many incomplete matrices, this paper proposes a matrix completion technique through generalized low rank approximations of matrices (GLRAM). Firstly, the matrix completion problem is formulated as a non-convex minimization model on the basis of each datum’s tri-factorization. Then, an alternating direction method (ADM) is employed to solving the proposed optimization problem. Finally, numerical results demonstrate that the proposed method has the attractive efficiency for recovering missing entries and improving compression ratio.

备注：

EI期刊检索

合写作者：

雍龙泉

第一作者：

史加荣

论文类型：

期刊论文

卷号：

卷：5

期号：

期：6

页面范围：

页：1619-1625

是否译文：

否

发表时间：

2014-12-01