一种基于快速分解模拟退火算法的数据聚类算法  

A Faster Decoposition-based Simulated Annealing Algorithm for the Data Clustering and Its Application

作　　者：张蓉[1] 彭宏[2] 

机构地区：[1]广东商学院信息学院,广州5103202 [2]华南理工大学计算机系,广州510640

出　　处：《计算机工程》2002年第8期88-89,110,共3页Computer Engineering

基　　金：广东省科委基金项目---(2000J00601)

摘　　要：把求解数据聚类问题转换为一个图形分割寻优问题,提出一种基于快速分解模拟退火算法的数据聚类算法。实验结果证明,快速分解模拟退火算法退火时间短,收敛速度快,磁盘开销小,把它应用于数据聚类中,可以获得较好的聚类结果。This paper formulates for solving the data clustering problem as a graph partition optimal problem , and proposes a method for clustering of data based on a faster decomposition-based simulated annealing algorithm . Both the analytical and experimental studies indicate that the faster decomposition-based simulated annealing algorithm can shorten annealing time ,quicken convergence speed and dramatically reduce the disk I/O activities , and when the algorithm has been used in data clustering , excellent clustering results can be obtained. ;;

关 键 词：快速分解模拟退火算法 数据聚类算法 数据库 数据挖掘 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]