正则化半监督判别分析方法

为了克服加权线性判别分析(WLDA)只利用有标签的训练样本而不能反映样本数据流形结构的缺点，提出一种正则化的半监督判别分析方法。首先构建所有样本的近邻图来估计数据的局部流形结构，然后将此作为正则项引入WLDA的准则函数中。该方法避免了类内散度矩阵奇异，同时保持了样本数据的判别结构和几何结构。在ORL和YALE人脸数据库上的实验结果证明了该算法的有效性。

Regularized Semi-supervised Discriminant Analysis for Face Recognition

A new semi-supervised discriminant analysis algorithm algorithm based on manifold regularization is proposed for the disadvantage of Weighted Linear Discriminant Analysis (WLDA). Which can avoid the singularity of the total-scatter matrix, and the discriminant structure and the intrinsic geometrical structure of the sample was be preserved. A nearest neighbor graph was constructed first to estimate the intrinsic geometrical structure of the sample, and then the graph structure was incorporated into the objective function of the multivariate linear regression as a regularization term. Experimental results on ORL and Yale face recognition demonstrate the effectiveness of the algorithm.