Price Sensitive Support Vector Machine Based on Data Set Decomposition

This paper presents a training set decomposition of price-sensitive support vector machines for classification Firstly, decomposition of the training sample set, each subset of training a support vector machine can output a posteriori probability, get through the training of support vector machine on the training sample posterior probability using the learning process and the cost matrix, the true class label of the sample, in order to achieve the reconstruction of the sample, contains a sample misclassification cost information, cost-sensitive support vector machines, making the classification of unbalanced data sets, so that the smallest misclassification cost.