摘要
特征选择是数据预处理的重要环节,可以提高数据质量,简化数据结构,降低后续环节计算成本。目前,许多研究将进化计算应用到特征选择领域,目的是求解数据集的一条最优特征子集。但是由于现实问题的复杂性,数据集中的最优特征子集存在不唯一的情况,并且用单个特征子集表征数据信息具有一定的片面性。因此,提出了一种基于多模态进化算法的特征选择策略,在对特征子集寻优的同时,定位多个最优及次优特征子集。最后,将本文算法与其他基于进化计算的特征选择策略在标准数据集中进行了对比,结果表明本算法具有一定的优越性。
Feature selection is an essential process in data preprocessing.It can improve the equality of dataset and save the computational costs.Many studies have applied evolutionary algorithms to the field of feature selection so as to find out the optimal feature subset.However,the optimal subset may not be unique due to the complexity of real problems.In addition,the single feature subset is one-sided and cannot represent the full information of the original data.Therefore,a feature selection strategy based on multimodal evolutionary algorithm is proposed to find out multiple optimal and suboptimal feature subsets while optimizing the feature subsets.Finally,the algorithm is compared with other feature selection strategies based on evolutionary computing in the standard data set,and the results show that the algorithm has certain advantages.
作者
朱小培
位云朋
闫李
韩茜茜
ZHU Xiaopei;WEI Yunpeng;YAN Li;HAN Xixi(School of Electronic and Information Engineering,Zhongyuan University of Technology,Zhengzhou 450007,China;School of Electrical Engineering,Zhengzhou University,Zhengzhou 450001,China)
出处
《中原工学院学报》
CAS
2021年第4期71-76,共6页
Journal of Zhongyuan University of Technology
基金
国家自然科学基金项目(61976237)
河南省科技攻关项目(212102210537)
中原工学院校内教改项目(JG202001)
关键词
特征选择
数据预处理
进化计算
多模态进化
feature selection
data preprocessing
evolutionary computation
multimodal evolutionary algorithm
