西安电子科技大学学报

• 研究论文 • 上一篇    下一篇

叠加信息熵游走数据聚类算法

续拓;李洁;王颖   

  1. (西安电子科技大学 电子工程学院,陕西 西安 710071)
  • 收稿日期:2017-11-03 出版日期:2018-08-20 发布日期:2018-09-25
  • 作者简介:续拓(1993-),男,西安电子科技大学硕士研究生,E-mail:2509870854@qq.com
  • 基金资助:

    国家自然科学基金资助项目(61432014,61172146,61201294);中央高校基本科研专项资金资助项目(JB140225);高等学校博士学科点专项科研基金资助项目(20120203120009,20121401120015)

Clustering by samples movement in the superposition information entropy field

XU Tuo;LI Jie;WANG Ying   

  1. (School of Electronic Engineering, Xidian Univ., Xian 710071, China)
  • Received:2017-11-03 Online:2018-08-20 Published:2018-09-25

摘要:

在数据聚类的过程中,由于样本数据空间分布的复杂性,相似度度量过程中的重复性以及算法的自适应性等问题,聚类算法往往无法得到正确的聚类结果.为了解决样本数据空间分布复杂的问题,提出叠加信息熵数据游走聚类算法.该算法通过在数值空间构建样本叠加信息熵场,并通过数据游走进行数据分割实现聚类.实验结果表明,该算法不仅可以获得较好的聚类效果,同时具有较高的数据自适应性.

关键词: 聚类, 信息熵, 数据游走

Abstract:

The essence of clustering is to compare the similarities among samples on different scales. Due to the complexity of the spatial distribution of sample data, the similarity measurement in the process of repeatability, and the algorithm for adaptive problems, the clustering algorithm cannot lead to the correct result in the process of data clustering. In order to solve the complex problem of spatial distribution of sample data, we present a data migration clustering algorithm based on superposition information entropy, which is used to construct the entropy field of the data in the numerical space, and the datawandering to implement data segmentation and complete clustering. Experimental results show that this method can not only obtain a better clustering effect, but also have data adaptability.

Key words: custering, information entropy field, samples movement