随机种子最近邻居搜索聚类算法研究

Study on random seed nearest neighbour search clustering algorithm

提出了随机种子最近邻居搜索(RS-NNS)聚类算法,该算法从随机确定的种子开始沿着它最近邻居的方向搜索具有最大相似特征的邻居对象,形成局部最大聚类集合,并在搜索过程中动态调整数据对象的归属,以实现局部的最优分配,直到所有的数据对象完成聚类标识。经过验证,该算法可以适应数据集合的密度、形状、噪音、聚类个数等问题,并且相对于同类算法可以实现较快地优化搜索。

This paper presents a random seed nearest neighbour search clustering algorithm （RS-NNS）. The method is to follow the nearest neighbours＇ direction of a random selected seed, search and find its neighhours which have the greatest similar features, form the local maximum cluster, adjust dynamically the data objects＇ belongingness to realize the local optimization, and end the clustering procedure until all the data objects are identified. Experiments verify that the new algorithm fits the problems such as different density, shape, noise, cluster number and so on, and can realize fast optimization searching.