Local k-proximal plane clustering
Local k-proximal plane clustering
复制标题
DOI：
10.1007/s00521-014-1707-9
复制
发表时间：
2014-09
影响因子：
6
通讯作者：
Zhimin Yang;Yan-Ru Guo;Chunna Li;Y. Shao
中科院分区：
文献类型：
--
作者：
Zhimin Yang;Yan-Ru Guo;Chunna Li;Y. Shao
k-平面聚类（kPC）和k-邻近平面聚类（kPPC）将数据点聚类到中心平面，而不是在k-均值中将数据点聚类到聚类中心。然而，kPC和kPPC构造的聚类中心平面是无限延伸的，这将影响聚类性能。在本文中，我们提出了一个局部k-邻近平面聚类（LkPPC），通过将k-means引入kPPC，这将迫使数据点集中在一些原型周围，从而局部化聚类中心平面的表示。LkPPC的主要贡献如下：（1）LkPPC引入了每个聚类中心平面的局部化表示，避免了无限混乱。(2)与kPPC不同的是，LkPPC构造了聚类中心平面，使得同一个聚类中的数据点既靠近中心平面又靠近原型，同时又远离其他聚类，从而解决了特征值问题。(3)代替随机选择初始数据点，建立拉普拉斯图策略来初始化数据点。(4)在多个人工数据集和基准数据集上的实验结果表明了LkPPC的有效性。
k-Plane clustering (kPC) and k-proximal plane clustering (kPPC) cluster data points to the center plane, instead of clustering data points to cluster center in k-means. However, the cluster center plane constructed by kPC and kPPC is infinitely extending, which will affect the clustering performance. In this paper, we propose a local k-proximal plane clustering (LkPPC) by bringing k-means into kPPC which will force the data points to center around some prototypes and thus localize the representations of the cluster center plane. The contributions of our LkPPC are as follows: (1) LkPPC introduces localized representation of each cluster center plane to avoid the infinitely confusion. (2) Different from kPPC, our LkPPC constructs cluster center plane that makes the data points of the same cluster close to both the same center plane and the prototype, and meanwhile far away from the other clusters to some extent, which leads to solve eigenvalue problems. (3) Instead of randomly selecting the initial data points, a Laplace graph strategy is established to initialize the data points. (4) The experimental results on several artificial datasets and benchmark datasets show the effectiveness of our LkPPC.