支持向量机在处理分类问题时,如果两类数据重叠严重会造成分类器过学习,降低泛化性能。为此提出了一种基于信息熵的数据修剪支持向量机EB-SVM(entropy based-support vector machine),其主要思想是通过计算样例信息熵删除部分边缘数据和边界处混淆程度较高的样例以及噪声数据,用较少的训练样例学习SVM分类器。实验结果表明,该方法能够有效提高SVM的泛化性能。 Support vector machine in dealing with classification problems, if two types of data overlap serious classifier will lead to learning, reduce generalization performance. Therefore, an information entropy-based data pruning support vector machine (EB-SVM) is proposed. The main idea is to delete part of the edge data and calculate the degree of confusion at the boundary by calculating the sample information entropy As well as the noise data, learn the SVM classifier with fewer training examples. Experimental results show that this method can effectively improve the generalization performance of SVM.