摘要
文中从实际工程需要出发,结合随机采样一致性原理,将每一个数据点描述为适合该点首选的假定模型的倾向集集合,再基于倾向集之间的Jaccard距离,采用一种经过改良的密集聚类方法,聚集属于同一种模型的数据点.这种方法既不要求事先指定模型数目,也不需要进行参数调试.实验表明,该算法检测效果明显,精确度较高.
Start from the actual needs of engineering, each data point is represented with a preference set of hypotheses models preferred by that point in this paper. Subsequently, based on random sample consensus principle, a improved agglomerative clustering should be adopted to collect those points which fall into a same model. The method does not require prior specification of the number of models, nor it necessitate parameters tuning. Experimental resuits demonstrate the obvious effect and greater accuracy of the algorithm.
出处
《微电子学与计算机》
CSCD
北大核心
2011年第11期194-196,200,共4页
Microelectronics & Computer
基金
国家"八六三"计划项目(2007AA01Z179)
关键词
倾向集
Jaccard距离
密集聚类
多圆
preference se
jaccard distance
agglomerative clustering
multiple-- circles
