摘要
给出了一种基于L2,1范数和局部约束的非负矩阵分解方法,降低了对原始数据中噪声和异常值的敏感程度,分析了该方法的严格收敛性,通过多个数据集上的实验结果验证该算法的有效性。
A non-negative matrix factorization method based on L2,1 norm and local constraints is presented,which reduces the sensitivity to noise and outliers in the original data.The strict convergence of this method is analyzed,and the effectiveness of this algorithm is verified by experimental results on several data sets.
作者
文学春
向远强
WEN Xuechun;XIANG Yuanqiang(School of Mathematical Sciences,Guizhou Normal University,Guizhou 550025,China)
出处
《新乡学院学报》
2023年第3期14-21,共8页
Journal of Xinxiang University
关键词
矩阵分解
L2
1范数
局部约束
行稀疏性
聚类
matrix decomposition
L2,1 norm
local constraints
row sparseness
cluster