《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (2): 375-381.DOI: 10.11772/j.issn.1001-9081.2021030383

• 人工智能 • 上一篇    

迭代直觉模糊K-modes算法

陈育丹, 高翠芳(), 沈莞蔷, 殷萍   

  1. 江南大学 理学院,江苏 无锡 214122
  • 收稿日期:2021-03-15 修回日期:2021-07-02 接受日期:2021-07-02 发布日期:2022-02-21 出版日期:2022-02-10
  • 通讯作者: 高翠芳
  • 作者简介:陈育丹(1998—),女,江西赣州人,硕士研究生,主要研究方向:计算智能、模式识别;
    高翠芳(1974—),女,河北石家庄人,副教授,博士,主要研究方向:模式识别、生物信息学;
    沈莞蔷(1981—),女,江苏常州人,副教授,博士,主要研究方向:计算机图形学、模式识别;
    殷萍(1981—),女,浙江嘉兴人,副教授,博士,主要研究方向:数值计算、模式识别。
  • 基金资助:
    国家自然科学基金资助项目(61772013)

Iterative intuitionistic fuzzy K-modes algorithm

Yudan CHEN, Cuifang GAO(), Wanqiang SHEN, Ping YIN   

  1. School of Science,Jiangnan University,Wuxi Jiangsu 214122,China
  • Received:2021-03-15 Revised:2021-07-02 Accepted:2021-07-02 Online:2022-02-21 Published:2022-02-10
  • Contact: Cuifang GAO
  • About author:CHEN Yudan, born in 1998, M. S. candidate. Her research interests include computational intelligence, pattern recognition.
    GAO Cuifang, born in 1974, Ph. D., associate professor. Her research interests include pattern recognition, bioinformatics.
    SHEN Wanqiang, born in 1981, Ph. D., associate professor. Her research interests include computer graphics, pattern recognition.
    YIN Ping, born in 1981, Ph. D., associate professor. Her research interests include numerical calculation, pattern recognition.
  • Supported by:
    National Natural Science Foundation of China(61772013)

摘要:

直觉模糊K-modes(IFKM)算法在聚类过程中采用简单0-1匹配相似性度量,既无法有效刻画类内数据对象之间的相似性,也未体现不同属性在聚类过程中的贡献程度;此外,IFKM算法在聚类的每一次迭代中直接根据直觉模糊隶属度矩阵来确定数据对象所属类别,没有充分发挥直觉模糊思想的作用。为了解决这两个问题,提出一种迭代IFKM (IIFKM)算法。首先,基于直觉模糊熵(IFE)与直觉模糊集(IFS)定义了一种加权的直觉模糊隶属度相似性度量;其次,将直觉模糊隶属度矩阵作为迭代信息贯穿于整个聚类过程,使算法中的直觉模糊思想得到充分体现。在UCI数据库的5个数据集上进行的实验结果表明,与IFKM算法相比,IIFKM算法在分类正确率和召回率方面提升了7%~11%,在分类精度方面也有一定提升。

关键词: 分类型数据聚类, 相似性度量, 直觉模糊K-modes算法, 直觉模糊集, 直觉模糊熵

Abstract:

Intuitionistic Fuzzy K-Modes (IFKM) algorithm adopts the simple 0-1 matching similarity measure in clustering process, which can not effectively describe the similarity of data objects in class, and fails to reflect the contribution of different attributes in clustering process. In addition, IFKM algorithm directly determines the classes of data objects according to the intuitionistic fuzzy membership matrix in each iteration of clustering, and do not give full play to the role of intuitionistic fuzziness idea. In order to solve these two problems, an Iterative IFKM (IIFKM) algorithm was proposed. Firstly, a weighted similarity measure of intuitionistic fuzzy membership degree was defined based on Intuitionistic Fuzzy Entropy(IFE) and Intuitionistic Fuzzy Set (IFS). Secondly, the intuitionistic fuzzy membership matrix was used as iterative information in the whole clustering process, so that the intuitionistic fuzziness idea in the algorithm was fully reflected. Experimental results on 5 datasets from UCI database show that compared with IFKM algorithm, the proposed IIFKM algorithm can improve the accuracy and recall by 7%-11%, and can also improve the precision to some degree.

Key words: categorical data clustering, similarity measure, Intuitionistic Fuzzy K-Modes (IFKM) algorithm, Intuitionistic Fuzzy Set (IFS), Intuitionistic Fuzzy Entropy (IFE)

中图分类号: