摘要
该算法(YNPF)主要是针对粒子群在优化聚类中心时运行时间过长而提出的。YNPF首先利用模糊聚类的有效性测量方法确定最佳聚类数目,然后,利用一种改进的粒子群优化(YNPSO)算法去优化模糊C均值(WAFCM[1])聚类的中心,最后,再用WAFCM进行聚类。试验表明,该算法能提高分类的正确率,提高运算速度,聚类效果优于使用基本的FCM、基本的PSO以及两者的简单结合(PF)和定标法[2]与WAFCM的结合(NPF)。
The algorithm （YNPF） is proposed because particle swarm optimizing with cluster centers spends a redundant time. Firstly YNPF uses a fuzz validity criterion to ensure the best number of clustering, secondly uses a improved particle swarm （PSO） algorithm to optimize the centers of fuzz C means （WAFCM[1]） clustering, lastly uses WAFCM again. The experiment result shows that the proposed algorithm can improve the classification correct rate, improve operation rapidity. The clustering effects are superior to original FCM algorithm or original PSO or hybrid clustering based on particle swarm optimization and FCM algorithm （PF） and NPF algorithm.
出处
《计算机安全》
2009年第2期33-35,共3页
Network & Computer Security
关键词
粒子群优化
模糊C均值聚类
梯度下降
群体智能
全局寻优
particle swarm optimization
fuzzy C means clustering
gradient drop
groups intelligence
global optimization
