摘要
聚类可以看成是寻找K个最佳聚类中心的过程。文中把一组聚类中心视为一个粒子(P),把各个数据到各自聚类中心的欧式距离之和看成优化函数(f(P)),使用带混沌搜索的粒子群聚类算法(C-PSO)算法寻找最优函数值,从而找到最佳聚类中心。该算法改进了粒子速度的初始化,把混沌搜索嵌入到粒子群的搜索过程中,提高了粒子群的搜索能力。实验结果表明,该算法的聚类效果明显好于K-means和PSO聚类。
Clustering can be regarded as the process of finding K optimal centers. Considered that a group of centers can be seen as a particle （P） ,and the sum of Euclidean distance between data and its clustering center as optimal function （f（P）） ,and then using particle swarm optimization clustering algorithm with chaos search to find the optimal function value,so as to find the optimal centers. This algorithm improved on initialization of particle velocity, embeding the chaos search into particle search, so improved the capability of global search of particle swarm. The experiment showed that the clustering result of this algorithm was better than K-means and PSO clustering.
出处
《计算机技术与发展》
2008年第10期93-95,102,共4页
Computer Technology and Development
