Semi-supervised dimensionality reduction based on linear local tangent space alignment and label propagation
Xue We · Jisuanji yingyong yanjiu · 2014
Considering the limit that linear local tangent space alignment( LLTSA) can't take advantage of the sample label information in face recognition application,this paper proposed a semi-supervised dimensionality reduction based on linear local tangent space alignment and label propagation( SSLLTSA). SSLLTSA used label propagation to get the soft labels in the sample data with part of labels. Then,it constructed the soft label based scatter matrices to describe the intra-class compactness and the inter-class separability. SSLLTSA used the information in the label effectively with the well preserved local structure of data. Through the experiments on YALE and ORL,SSLLTSA outperforms based on traditional dimensionality reduction algorithms with maximum average recognition rate by 3. 50% and 3. 89% respectively.