Feature selection for support vector machines

In the context of support vector machines (SVM), high dimensional input vectors often reduce the computational efficiency and significantly slow down the classification pro-cess. In this paper, we propose a strategy to rank individual components according to their influence on the class assign-ments. This ranking is used to select an appropriate subset of the features. It replaces the original feature set without significant loss in classification accuracy. Often, the gen-eralization ability of the classifier even increases due to the implicit regularization achieved by feature pruning. 1