摘要
为实现通过提高癌前病变分类准确率,以降低结直肠癌的发生率和死亡率,提出一种基于MMTS-AdaBoost的高维数据分类算法,优化高维数据分类算法,提高分类性能。通过将本征正交思想引入马田系统,构建改进马田系统获取重要特征变量实现降维。使用降维得到的特征,应用AdaBoost算法对癌前病变类型进行分类。实验结果表明,与使用降维处理的mrmr-AdaBoost和chisquare-AdaBoost算法,以及AdaBoost、BP网络、NB、SVM等经典分类算法相比,MMTS-AdaBoost的F1和G-mean更高,分类性能更优。
In order to reduce the incidence and mortality of colorectal cancer by improving the classification accuracy of precancerous lesions,a high-dimensional data classification algorithm based on MMTS-AdaBoost is proposed to optimize the classification algorithm of high-dimensional data and improve the classification performance.The dimension reduction was achieved by applying the proper orthogonal decomposition idea to Mahalanobis-Taguchi system and improving the Mahalanobis-Taguchi system to acquire important feature variables.Using the feature variables obtained by dimension reduction,the AdaBoost algorithm was used to classify the types of precancerous lesions.The experimental results show that compared with the mrmr-AdaBoost and chisquare-AdaBoost algorithms that use dimension reduction,as well as the classic classification algorithms such as AdaBoost,BP network,NB,and SVM,the F1 and G-mean of the MMTS-AdaBoost are higher,and the classification performance is better.
作者
茅婷
张月义
孙叶芳
虞岚婷
Mao Ting;Zhang Yueyi;Sun Yefang;Yu Lanting(School of Economics and Management,China Jiliang University,Hangzhou 310000,Zhejiang,China)
出处
《计算机应用与软件》
北大核心
2024年第1期291-296,共6页
Computer Applications and Software
基金
国家社会科学基金项目(18BJY033)。
关键词
结直肠癌癌前病变
高维数据分类
马田系统
ADABOOST
本征正交分解
Precancerous lesions of colorectal cancer
High-dimensional data classification
Mahalanobis-Taguchi system
AdaBoost
Proper orthogonal decomposition
