摘要
数据挖掘技术在中医辅助诊断中被日益重视,计算机辅助诊断本质上是一个数据挖掘分类任务。针对中医临床数据的模糊性和不完整性,提出了一种基于特征提取的分类集成模型。这种模型能将扰动训练数据和扰动输入属性结合起来,生成精确且差异度大的个体分类器。与单个分类器和其他集成方法的对比实验,证明这种新模型在脾虚证辅助诊断上有更低的错误率。进一步的实验显示特征提取在这种新模型中对降低错误率有显著的作用。
Data mining attracts increasing attention in computer-aided diagnostic system of traditional Chinese medicine. In essential, computer-aided diagnosis is a classification task. The clinical data of traditional Chinese medicine is characterized by incompletion and ambiguity. According to this feature, in this paper we propose a new classification ensemble model which is based on feature selection. This model is able to generate precise individual classifier with big diversity by combining the disarrangements of training data and input attributes. The contrast experiments carried on with the single classifier and other integration methods prove that this new model has lower error rate in computer-aided diagnosis of Spleen asthenia. Further experiments indicate that the feature selection exerts an obviously positive effect in minimizing error rate.
出处
《计算机应用与软件》
CSCD
2010年第3期22-25,共4页
Computer Applications and Software
基金
国家自然科学基金重大研究计划重点项目(90209004)
关键词
分类集成
特征提取
计算机辅助诊断
脾虚证
Classification ensemble Feature selection Computer-aided diagnosis Spleen asthenia
