| 黄元江,汤德佑,胡红武.动态迭代聚类算法分析基因序列数据[J].南华大学学报(自然科学版),2004,(4):57~60.[.Analyzing Gene Sequence Data by Dynamically Iterative Clustering Algorithms[J].Journal of University of South China(Science and Technology),2004,(4):57~60.] |
| 动态迭代聚类算法分析基因序列数据 |
| Analyzing Gene Sequence Data by Dynamically Iterative Clustering Algorithms |
| 修订日期：2004-10-08 |
| DOI： |
| 中文关键词: 动态迭代聚类算法 基因数据库 知识发现 K—均值算法 基因序列 海量序列 |
| 英文关键词:Clustering,Gene database,knowledge discovery,K-Means Clustering Algorithms |
| 基金项目: |
| 黄元江 汤德佑 胡红武 |
| [1]株洲工学院计算机系,湖南株洲,412008 [2]华南理工大学软件学院,广东广州,510630 [3]株洲工学院计算机系,湖南株洲,412008／／华南理工大学软件学院,广东广州,510630 |
| 摘要点击次数: 984 |
| 全文下载次数: 1 |
| 中文摘要: |
| 聚类技术在知识发现方面发挥了很重要的作用，K—均值算法是聚类分析中最常用的算法，但K—均值算法必须预先选择类的数目作为先验值，即研究者需要确定数据空间内有意义类的数目．针对这个问题，本文提出一种新的聚类算法—动态迭代聚类算法，动态选取K个边缘相似度的数据对象作为最初的初始聚类点，并根据类内或类间的相似度离差程度不断地精练(合并或分割)初始类群．模拟实验结果表明，该算法提高了聚类质量，使聚类具有更高的准确性。 |
| 英文摘要: |
| Clustering technology is very important in knowledge discovery,and K-Means Clustering Algorithms is the most frequently used in clustering analysis.But K-Means Algorithms must choose the amount of classes in advance,in other words,researchers need the exact number of significant classes in data spaces.In this paper,we propose a new Clustering Algorithm aiming at the matter mentioned above.The algorithm selects dynamically K marginal similar objects as original class points and refines constantly these class points based on the discrete degree inner classes or between classes.The simulation results show that this Algorithm has improved veracity and efficiency,compared to the original algorithm. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|