Search

In this paper, the linear support vector machine (LPSVM) algorithm is used to construct an over complete set of weak classifiers, and AdaBoost algorithm are adopted to select part of them to form a strong classifier. During the course of feature extraction and selection, the new method can minimize the classification error directly, whereas most previous works cannot do this. An important difference between this method and other methods is that the sparse features are learnt from the training set instead of being arbitrarily defined. Experiments demonstrate that the new algorithm performs well.

To overcome the vast computation of the standard support vector machines (SVMs), Lee and Mangasarian (see First SIAM International Conference on Data Mining, 2001) proposed reduced support vector machines (RSVM). But they select 'support vectors' randomly from the training set, and this will affect the test result. In this paper, we select some representative vectors as support vectors via a simple unsupervised clustering algorithm, and then apply the RSVM method on these vectors. The proposed method can get higher recognition accuracy with fewer support vectors compared to the original RSVM, with the advantage of reducing the running time significantly.