摘要
在基于机器学习的医疗诊断系统中,分类算法的设计至关重要。为了提高医疗诊断系统的分类准确率,提出了先降维后分类的方法。采用有监督的LLE算法对高维医学数据进行特征提取。通过SVM算法对降维后的医学数据进行分类。以UCI数据库为数据来源,在MATLAB平台上进行各种分类算法的比较。实验结果表明,新算法的分类识别率和时间复杂度均优于传统的分类算法,非常适用于医学诊断领域。
In the machine learning based medical diagnosis system, the design of classifier is very important. In order to improvethe classification accuracy of the medical diagnosis system, a method is proposed that reduces the dimension firstly and then dothe classification. Supervised locally linear embedding method is used to feature extraction of high dimensional medical data.Support vector machine is used to classify the data after dimensionality reduction. Using UCI database as data source, variousclassification algorithms are compared on MATLAB platform. The experimental results show that the new method has higherclassification recognition rate and shorter running time than the traditional classification algorithms.
作者
张蕾
何萍
荣静
Zhang Lei;He Ping;Rong Jing(Guangling College,Yangzhou University,Yangzhou,Jiangsu 225000,China;College of Information Engineering,Yangzhou University)
出处
《计算机时代》
2018年第11期60-63,共4页
Computer Era
关键词
医学诊断
有监督的LLE算法
SVM算法
medical diagnosis
supervised locally linear embedding
support vector machine
