当前位置: 首页 > 文章 > 小样本数据的MIFS过滤式特征选择算法 山东农业大学学报(自然科学版) 2019,50 (1) 145-149
Position: Home > Articles > MIFS Filtering Feature Selection Algorithm for Small Sample Data Journal of Shandong Agricultural University(Natural Science Edition) 2019,50 (1) 145-149

小样本数据的MIFS过滤式特征选择算法

作  者:
王波;李时辉;郑鹏飞
单  位:
义乌工商职业技术学院
关键词:
小样本;MIFS;算法
摘  要:
针对小样本数据特征选择以及最佳特征难确定的问题,本文提出一种MIFS过滤式特征选择算法,同时结合Boruta算法,旨在降低数据集维度,确定出最佳特征的子集。通过实验结果与分析,对比其它三种传统的过滤式算法,验证本文算法的有效性。结果表明:MIFS-Boruta算法体现出更广的特征选择量,并且平均最低分类错误率最低。
译  名:
MIFS Filtering Feature Selection Algorithm for Small Sample Data
作  者:
WANG Bo;LI Shi-hui;ZHENG Peng-fei;Yiwu Industrial & Commercial College;
关键词:
Small sample;;NIFS;;algorithm
摘  要:
Aiming at the problem of feature selection for small sample data and the difficulty of determining the best feature,a MIFS filtering feature selection algorithm was proposed, which combined Boruta algorithm to reduce the dimension of data set and determine the subset of the best feature. Through the experimental results and analysis, comparing with the other three traditional filtering algorithms, the effectiveness of the proposed algorithm was verified. The results showed that MIFS-Boruta algorithm had broader feature selection and the lowest average classification error rate.

相似文章

计量
文章访问数: 10
HTML全文浏览量: 0
PDF下载量: 0

所属期刊

推荐期刊