摘要
大数据背景下,数据量呈指数级增长,三支决策在处理代价敏感问题时动态机制和稳定性不足.针对这个问题,结合F-粗糙集处理动态数据方面的优势,在代价敏感决策表簇中提出基于F-粗糙集和三支决策的平均代价敏感并行约简.首先,从平均决策代价和平均测试代价的角度,定义基于F-粗糙集和三支决策的并行约简;其次,设计基于F-粗糙集和三支决策的平均代价敏感并行约简算法.与基于分类的最小代价约简和基于类特定的最小代价约简比较,实验结果显示,基于F-粗糙集和三支决策的平均代价敏感并行约简可以更好地权衡误分类代价(决策代价)和测试代价,提高分类准确率.研究结果为研究动态决策和代价敏感提供一种新的研究方法和思路.
In the era of big data,data increased exponentially,the existed three-way decisions became insufficient of dynamic mechanism and stability to deal with cost-sensitive issues.In order to compensate the insufficiency,an average cost-sensitive parallel reducts based on F-rough sets and three-way decisions was proposed to deal with cost-sensitive information in a family of decision systems.Firstly,combined with both average decision cost and average test cost,the parallel reducts based on F-rough sets and three-way decisions were defined.Secondly,an average cost-sensitive parallel reduct algorithm based on F-rough sets and three-way decisions was developed.Compared with two algorithms(the classification-based minimum cost reduction algorithm and the class-specific minimum cost reduction algorithm),experimental results showed that the average cost-sensitive parallel reducts based on F-rough sets and three-way decisions could better balance the misclassification cost(decision cost) and the test cost of decision classes,and could also improve the classification accuracy.The presented approaches also suggested a new direction for further study of dynamic decision making and cost sensitivity.
作者
邓大勇
刘月铮
肖春水
DENG Dayong;LIU Yuezheng;XIAO Chunshui(Xingzhi College,Zhejiang Normal University,Lanxi 321100,China;College of Mathematics and Computer Science,Zhejiang Normal University,Jinhua 321004,China;Key Laboratory of Intelligent Education Technology and Application of Zhejiang Province,Zhejiang Normal University,Jinhua 321004,China)
出处
《浙江师范大学学报（自然科学版）》
CAS
2023年第1期7-17,共11页
Journal of Zhejiang Normal University:Natural Sciences
基金
浙江省科技计划资助项目(2020C35066)。
关键词
代价敏感
三支决策
属性约简
F-粗糙集
cost sensitivity
three-way decisions
attribute reduction
F-rough sets
