摘要
数据特征选择就是从初始的数据特征中选择指定数据进行子集筛选。目前,通常使用人工蜂群算法进行特征选择,但由于收敛慢、寻优差,无法满足人们的需求。因此,本文提出一种改进人工蜂群算法,通过特征选择绘制大数据特征选择框架图,建立多项搜索渠道;利用改进的人工蜂群算法提取并行特征,使用MapReduce模型降低编程难度,获取并行特征最优解;设计特征选择复杂粗糙集模型,并构建特征学习模型来实现大数据特征选择。试验结果表明,设计的特征选择方法性能优于传统方法。
Data feature selection is to select specified data from the initial data features for subset filtering.Currently,artificial bee colony algorithms are usually used for feature selection,but due to slow convergence and poor optimization,it cannot meet people′s needs.Therefore,this paper proposes an improved artificial bee colony algorithm,which draws the framework of big data feature selection through feature selection,and establishes multiple search channels;uses an improved artificial bee colony algorithm to extract parallel features,uses the MapReduce model to reduce programming difficulty,and obtains the optimal solution for parallel features;designs a complex rough set model for feature selection,and builds a feature learning model to realize big data feature selection.The test results show that the performance of the feature selection method designed in this paper is better than the traditional method.
作者
李玮瑶
LI Weiyao(Computer School,Pingdingshan University,Pingdingshan Henan 467000)
出处
《河南科技》
2021年第19期27-29,共3页
Henan Science and Technology
关键词
改进人工蜂群算法
大数据
特征选择
improve artificial bee colony algorithm
big data
feature selection
