吉林大学学报(信息科学版)

• 论文 • 上一篇    下一篇

DBSCAN算法在高性能计算中心用户分类的应用研究

徐海啸a,b, 麻婧a, 吴旗a,b   

  1. 吉林大学 a. 计算机科学与技术学院; b. 高性能计算中心, 长春 130012
  • 收稿日期:2013-02-19 出版日期:2013-09-24 发布日期:2014-04-04
  • 作者简介:徐海啸(1985—), 男, 长春人, 吉林大学工程师, 主要从事高性能计算研究, (Tel)86-0431-85155256(E-mail)haxxiao@jlu.edu.cn;通讯作者: 麻婧(1990—), 女, 黑龙江齐齐哈尔人, 吉林大学本科生, 主要从事数据挖掘研究, (Tel)86-13944859775(E-mail)hamazuhe2009@163.com。
  • 基金资助:

    大学生创新实验国家级基金资助项目(2011A53101)

Application Research of DBSCAN Algorithm Based on High-Performance Computing Center Users Classification

XU Hai-xiaoa,b, MA Jinga, WU Qia,b   

  1. a. College of Computer Science and Technology; b. High Performance Computing Center, Jilin University, Changchun 130012, China
  • Received:2013-02-19 Online:2013-09-24 Published:2014-04-04

摘要:

        为提高集群资源使用效率, 管理员需要对用户进行分类, 从而对不同用户提出资源使用策略。DBSCAN(Density Based Spatial Clustering of Applications with Noise)聚类算法可对用户进行分类, 但对初始参数敏感。为此, 提出改进算法, 首先将密度进行层次划分, 由此得出各层次的密度阈值, 在每种阈值下采用DBSCAN算法, 解决全局参数问题。在此基础上, 创新地使用一个直接可达距离排序队列, 将排序信息作为可变参数, 减小初始参数对结果的影响。通过高性能计算中心用户数据的实例验证了其可行性。实验结果表明, 改进后的算法提高了用户分类的准确性和全面性。

关键词: 聚类分析, DBSCAN算法, 高性能计算中心, 用户分类, 数据挖掘

Abstract:

To enhance service efficiency on cluster resource, administrator needs to make classification of users, and provide various strategies on resource utilization to different users. DBSCAN (Density Based Spatial Clustering of Applications with Noise) algorithm can achieve users classification, but the initial parameters are very sensitive. The improved algorithm classifies the level of density firstly, then gets the densitythreshold of each level, and uses DBSCAN under each threshold which solves the problem of global  parameters. It uses a sorted queue of directly accessible distance as an innovation, makes the sorting information as variable parameter to decrease the influence of initial parameter. The algorithm has verified its feasibility through example data of HPC users. The experimental result demonstrates that this improved algorithm can achieve a more accurate and comprehensive user classification.

Key words: clustering analysis, density based spatial clustering of applications with noise (DBSCAN), high-performance computing center, users classification, data mining

中图分类号: 

  • TP391