The use of mass spectrometry as a proteomics tool is poised to revolutionize early disease diagnosis and treat. Unfortunately

due to existing “the curse of dimensionality”

most standard machine learning techniques cannot be directly applied to recognition proteomic mass spectrometry

and these methods also faced with the problem of poor recognition performance. For better efficiency to use principal component analysis (PCA) and local linear discriminant embedding (LLDE) for face recognition

a Double-Time Projections method and a Modified Double-Time Projections method are proposed for recognition proteomic mass spectrometry. The proposed methods first do denoising and dimension reduction by T-test

and then obtain first projection feature vectors with minimum mean square error and get second projection feature vectors with maximum separability. Finally

preprocessed data are projected in the sub space based on two feature vectors and

are classified. The experimental result from the dataset of ovarian cancer proteomic mass spectrometry indicates that the proposed methods have a better accuracy than available methods.