【摘要】 主要针对传统的聚类算法倾向于识别大小类似的球形聚类簇,且对离群数据较为敏感等问题,利用聚类簇代表点选取的方法,同时结合基于人进行聚类判断所遵循的基本原则,即聚类中对象间距离应小于聚类间距离,设计了一种有效的聚类算法,实验结果表明算法是有效的。更多还原

【Abstract】 Based on the method of selecting the representa ti ve points of clustering clusters and the principle that the distance between obj ects in clusters must be shorter than that between clusters when judging cluster s, this paper focuses on problem that the algorithm of traditional clustering is sensitive to the independent data and inclined to recognize the spheral cluster ing clusters which are similar in size, and designs an effective clustering algo rithm. The result indicates that this algorithm is effective when handling compl icated data.更多还原