电子学报 ›› 2018, Vol. 46 ›› Issue (2): 289-297.DOI: 10.3969/j.issn.0372-2112.2018.02.005

• 学术论文 • 上一篇    下一篇

MapReduce框架下的粒概念认知学习系统研究

米允龙1,2, 李金海1, 刘文奇1, 林晶2,3   

  1. 1. 昆明理工大学理学院, 云南昆明 650500;
    2. 怀化学院计算机科学与工程学院, 湖南怀化 418000;
    3. 武陵山片区生态农业智能控制技术湖南省重点实验室, 湖南怀化 418000
  • 收稿日期:2016-09-26 修回日期:2016-11-17 出版日期:2018-02-25 发布日期:2018-02-25
  • 通讯作者: 李金海
  • 作者简介:米允龙,男,1987年生于湖南怀化.硕士研究生,讲师,主要研究方向为数据挖掘与认知计算.E-mail:yunlongmi@yeah.net
  • 基金资助:
    国家自然科学基金(No.61305057,No.61562050,No.61573173)

Research on Granular Concept Cognitive Learning System Under MapReduce Framework

MI Yun-long1,2, LI Jin-hai1, LIU Wen-qi1, LIN Jing2,3   

  1. 1. Faculty of Science, Kunming University of Science and Technology, Kunming, Yunnan 650500, China;
    2. College of Computer Science and Engineering, Huaihua University, Huaihua, Hunan 418000, China;
    3. Key Laboratory of Intelligent Control Technology for Wuling-Mountain Ecological Agriculture in Hunan Province, Huaihua, Hunan 418000, China
  • Received:2016-09-26 Revised:2016-11-17 Online:2018-02-25 Published:2018-02-25

摘要: 针对经典的概念学习算法难以处理大规模数据集的问题,本文提出一种基于MapReduce框架的粒概念认知学习并行算法.该算法借鉴认知心理学的知觉和注意认知思想,并融合粒计算的粒转移原理.首先构建适应大数据环境的粒概念并行求解算法,并与经典粒概念构造算法做了对比,在此基础上分别从外延和内涵角度建立了粒概念认知计算系统,然后对给定对象集或属性集进行认知概念学习.实验结果表明,该并行算法是有效的,适合海量数据的粒概念认知学习.

关键词: 概念格, 概念学习, 认知计算, 粒计算, MapReduce

Abstract: Considering that the classical concept learning algorithms are difficult to deal with the massive data set,a MapReduce-based parallel algorithm for granular concept cognitive learning is proposed.The parallel algorithm is based on the cognitive thoughts of perception and attention in cognitive psychology,and it is combined with the granule transformation principle of granular computing.Specifically,a parallel algorithm is developed to compute granular concepts in big data environment,and a comparative analysis of the parallel algorithm and the classical granular concept construction algorithm is made.Granular concept cognitive computing systems are also constructed from the perspectives of extension and intension.Then,cognitive concept learning is performed by a given object set or attribute set.Experimental results show that the proposed parallel algorithm is effective and can be suitable for granular concept cognitive learning of massive data.

Key words: concept lattice, concept learning, cognitive computing, granular computing, MapReduce

中图分类号: