We tested the performance of LPP with different percent information kept in the PCA step. We tried options.PCARatio (= 1, 0.99, 0.98, 0.97). ==================================== options = []; options.PCARatio = 1 (0.99/0.98/0.97); [Vec,Val]=LPP(fea_Train,W,options); %================================= The affinity matrix is constructed as follow: %==================================== options = []; options.Metric = 'Cosine'; options.NeighborMode = 'Supervised'; options.WeightMode = 'Cosine'; options.bSelfConnected = 1; options.gnd = gnd_Train; W = constructW(fea_Train,options); %================================= The data is used without further pre-processing.
The only difference of the experimental results shown on the right from previous one is that we pre-process the data by normalizing each face vector to the unit.