代价敏感支持向量机

以分类精度为目标的传统分类算法通常假定：每个样本的误分类具有同样的代价且每类样本数大致相等．但现实数据挖掘中该假定不成立时，这些算法的直接应用不能取得理想的分类和预测．针对此缺隙，并基于标准的SVM，通过在SVM的设计中集成样本的不同误分类代价，提出代价敏感支持向量机（CS-SVM）的设计方法．实验结果表明CS-SVM是有效的．

Cost Sensitive Support Vector Machines

Classical methods of designing classifier generally pursue more highly accuracy based on the assumption that all misclassifications have the same cost and the sample number of each class is approximately equal.However,the assumption is not valid in some real applications such as fraud detection and medical diagnosis,so that classification algorithms without taking different misclassification cost into account do not perform well.Based on standard support vector machines(SVM),the algorithm of cost-sensitive SVM(CS-SVM) is proposed by integrating misclassification cost of each sample into standard SVM.Experimental results show that CS-SVM is effective.