论文部分内容阅读
信息熵是用来描述和度量事件发生不确定度的一种方法,能够把一些模糊量进行合理量化。本文利用分类对象样本的重要程度建立样本概率空间,把信息熵作为调节到支持向量机(SVM)核函数权重的依据,提出了基于熵权核函数的支持向量机方法。该方法首先利用信息交互熵计算各个特征对分类任务的重要度,然后用熵函数对样本的重要度度量核函数中的内积和欧氏距离,从而更加有效的支配核函数。理论分析和数值实验的结果都表明,该方法比传统的SVM具有更有效的分类能力和范化能力。
Information entropy is a method used to describe and measure the uncertainty of events, which can quantify some fuzzy quantities rationally. In this paper, we establish the sample probability space based on the degree of importance of the taxonomic samples, and use information entropy as the basis for adjusting the weight of SVM kernel function. We propose a support vector machine method based on entropy weight kernel function. In this method, we first calculate the importance of each feature to the classification task by using the information exchange entropy, and then use the entropy function to measure the inner product and the Euclidean distance of the kernel in the importance of the sample, so as to control the kernel function more effectively. The results of both theoretical and numerical experiments show that this method has more effective classification and normalization than the traditional SVM.