摘要
根据模拟退火具有寻求全局最优解的特性,在分析模拟退火基本理论的基础上,利用模拟退火基本思想对传统的空间聚类方法——K-means算法进行优化。然后分别对优化后的算法和传统算法进行实验分析。实验结果表明:优化后的方法以概率接受劣解的方式跳出局部极小值,从而为寻求全局的最优解提供了可能。另外,在优化过程中提出并应用了点密度的思想,使得聚类结果不受初始值影响,其执行效率也有所提高。
Simulated annealing has the feature of seeking the best solution.After analyzing the basic theories of simulated annealing,the traditional spatial clustering method K-means algorithm was optimized.Compared the optimized K-means algorithm with the traditional method by examples,the following conclusion was got that the optimized method dropped the local minimum by means of accepting inferior solution with probability,thus it could provide the possibility of seeking overall optimal solutions.Furthermore,it proposed and applied the idea of point density in the process of optimization,which made the clustering results unaffected by the initial value and promoted its efficiency as well.
出处
《测绘科学技术学报》
北大核心
2010年第4期306-309,共4页
Journal of Geomatics Science and Technology
基金
国家自然科学基金资助项目(40701157
40620130438
40671162)
国家863计划资助项目(2007AA12Z211)
关键词
模拟退火
空间聚类
K-MEANS算法
聚类准则函数
点密度
simulated annealing
spatial clustering
K-means algorithm
clustering criterion function
point density
