Local preserving projections andwithin-class scatter based semi-supervised support vector machines

The support vector machines (SVMs), as one of special regularization methods, has been used successfully in the field of pattern recognition. However, the traditional SVMs, a supervised learning method, gets the normal vector of the decision boundary mainly according to the largest interval law but has not taken the underlying geometric structure and the discriminant information into full consideration. Therefore, a local preserving projection and within-class scatter based semi-supervised support vector machine: LWSSVM, is presented in this paper by incorporating the basic theories of the locality preserving projections (LPP) and the linear discriminant analysis (LDA) in the SVMs. This method inherits the characteristics of the traditional SVMs, fully considers the global and local geometric structure between the samples and shows the global and local underlying discriminant information so that the classification accuracy can be increased. The tests on the face recgonition datasets show the above mentioned advantages of the LWSSVM method.