一种能发现自然聚类的聚类算法 *

目前的聚类算法如K-means、DBSCAN等，采用全局参数而难以发现数据的自然聚类，提出一种新的分 级聚类算法CluFNC，能够在数据空间中发现内部聚类特征。该算法的参数包括网格大小、噪声阈值和神经节点 数量。算法首先根据参数对数据空间划分网格，接着使用高斯影响函数计算每个单元的场强，然后针对网格位 置和网格的场强使用SOM进行聚类，最后使用Chameleon算法对SOM聚类得到的神经网络节点的权值进行聚 类，并把聚类结果映射回原始数据空间以得到最终聚类结果。理论分析和实验结果证明，该算法能够发

Clustering approach of finding natural cluster

Existing clustering algorithms,such as K-means,DBSCAN and so on,use global parameters and are difficult to find the natural cluster.This paper provided a novel cluster algorithm called CluFNC which could describe characters of intrinsic clustering,it had three parameters: length of the grid,threshold of noise and the count of neural node.The algorithm first created grid according to the length of grid and maps all data to the grid.Second,used Gausses function to calculate the power of every unit.Third,used ...