基于动态隧道系统的K-means聚类算法研究

针对K-means聚类算法易陷入局部极小的问题,利用动态隧道算法在解决全局最优化问题中的有效性,将算法中的动态隧道过程引入到K-means聚类算法中,提出了一种基于动态隧道算法的K-means聚类算法.该算法在K-means聚类算法寻优得到的局部极小值基础上,利用动态隧道过程寻找更小的能量盆地,再将其值提交给K-means聚类算法进行迭代寻优,重复该过程,直到找到全局最小值.理论分析和仿真实验证明,该算法的聚类效果要优于K-means聚类算法.

Research into K-means Clustering Algorithm Based on Dynamic Tunneling System

K-means clustering algorithm itself is a global optimization problem,whose objective function has multiple local minima and only one global minimum.The algorithm is apt to fall into local minimum points through its iteration process.Dynamic tunneling algorithm is especially developed for the global optimization problem.On the basis of the local minimum got by dynamic optimization process,dynamic tunneling process makes use of tunneling to skip the local minimum point and search a lower energy valley that in...