A fast training algorithm for support vector machine via boundary sample selection

A fast training algorithm based on boundary sample selection is proposed for support vector machine (BSS-SVM). This novel algorithm selects boundary samples from training set by fuzzy C-means clustering (FCM) algorithm to train SVM, instead of using normal training samples. Thus the scale of the training set is reduced greatly and the training speed of SVM is improved enormously. Experimental results show that the training speed of BSS-SVM is much faster than traditional algorithms without lose of any precision, especially for large training set.