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。I/O

关键词: 数据聚类, 图形分割, 模拟退火算法

Abstract: 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. ;;

中图分类号: 

• TP311.12