稀疏约束图正则非负矩阵分解

摘要

非负矩阵分解(NMF)是在矩阵非负约束下的一种局部特征提取算法.为了提高识别率,提出了稀疏约束图正则非负矩阵分解方法.该方法不仅考虑数据的几何信息,而且对系数矩阵进行稀疏约束,并将它们整合于单个目标函数中.构造了一个有效的乘积更新算法,并且在理论上证明了该算法的收敛性.在ORL和MIT-CBCL人脸数据库上的实验表明了该算法的有效性.%Nonnegative matrix factorization(NMF) is based on part feature extraction algorithm which adds nonnegative constraint into matrix factorization. A method called graph regularized non-negative matrix factorization with sparseness constraints (GNMFSC) was proposed for enhancing the classification accuracy. It not only considers the geometric structure in the data representation, but also introduces sparseness constraint to coefficient matrix and integrates them into one single objective function. An efficient multiplicative updating procedure was produced along with its theoretic justification of the algorithmic convergence. Experiments on ORL and MIT-CBCL face recognition databases demonstrate the effectiveness of the proposed method.