基于支持向量机与粗糙集理论的病例诊断

Diagnosis Cases based on Support Vector Machine and Rough Set

针对支持向量分类机在病例诊断中,训练样本大、诊断速度慢的不足,根据粗糙集理论的属性约简和支持向量机的分类机理,提出了一种混合分类算法,对病例进行诊断.应用粗糙集理论在不损失有效信息的情况下对属性进行预处理,从决策表中删除冗余的属性和冲突对象,降低支持向量机的维数和分类过程中的复杂度.然后利用支持向量机的分类机原理,对对象进行分类和预测,从而达到对病例进行诊断.实验证明在通过粗糙集对信息约简后,在合理降低准确率的情况下提高了诊断速度,从而解决了支持向量分类机在处理大量病例信息情况下,诊断速度慢的问题.

In order to solve the problem of lots of training data and slow speed in diagnoses of diseases based on SVM, a hybrid classification algorithm is put forward on the basis of attribute reduction of RS and classification principles of SVM to diagnose diseases. This algorithm firstly uses attribute reduction of RS to delete redundant attributes and conflicting objects from decision making table,without the loss of valued informations to reduce the dimension of SVM and the complexity in the process of SVM classification. Then this algorithm uses classification principles of SVM to classify and forecast the objects to achieve the effective diagnosis of diseases. This algorithm is proved through experiment that it really improves correction rate of diagnosis under the condition of reasonable reducing accurate rate and increasing the slow speed of diagnosing diseases while SVM dealing with much disease information.