Semi-supervised LLE algorithm of face recognition
Hongbing Zhou · Jisuanji gongcheng yu sheji · 2011
In practical applications,its’ easy to collect a small amount of labeled faces data sets.In order to make full use of the semi information of faces data sets.Based on the theory of density cluster,a new semi-supervised LLE algorithm is proposed.First,using every labeled sample as central point,then the label extension arithmetic is utilized to add labels to the unlabeled samples.Then,running supervised LLE Arithmetic,so we can get the dimension reduction data and the dimension reduction data contain the classified information of the original data sets.At last,the arithmetic is used to face recognition.Through comparing with the 2DLDA,LLE,SLLE,the arithmetic is proved effective.