聚类分析
计算机科学
光谱聚类
算法
模糊聚类
相关聚类
k均值聚类
CURE数据聚类算法
k-中位数聚类
树冠聚类算法
模式识别(心理学)
单连锁聚类
数据挖掘
作者
Guangchun Chen,Juan Hu,Hong Peng,Wang Jun,Xiangnian Huang
标识
DOI:10.15837/ijccc.2018.5.3238
摘要
Using spectral clustering algorithm is diffcult to find the clusters in the cases that dataset has a large difference in density and its clustering effect depends on the selection of initial centers. To overcome the shortcomings, we propose a novel spectral clustering algorithm based on membrane computing framework, called MSC algorithm, whose idea is to use membrane clustering algorithm to realize the clustering component in spectral clustering. A tissue-like P system is used as its computing framework, where each object in cells denotes a set of cluster centers and velocity-location model is used as the evolution rules. Under the control of evolutioncommunication mechanism, the tissue-like P system can obtain a good clustering partition for each dataset. The proposed spectral clustering algorithm is evaluated on three artiffcial datasets and ten UCI datasets, and it is further compared with classical spectral clustering algorithms. The comparison results demonstrate the advantage of the proposed spectral clustering algorithm.
科研通智能强力驱动
Strongly Powered by AbleSci AI