摘要
针对传统的模糊C-均值(FCM)聚类算法的聚类有效性对空间样本分布的依赖性等缺点,提出了一种新的基于人工鱼群算法的动态模糊聚类。通过引入模糊等价矩阵来表示高维样本之间的相似程度,并将高维样本映射到二维平面。然后利用人工鱼群算法不断优化二维样本的坐标值,使样本之间的欧氏距离向样本间的模糊等价矩阵趋近,最终实现模糊聚类。该方法克服了聚类有效性对高维样本空间分布的依赖性并同时提高了效率。仿真实验结果证明了该算法是有效的,具有聚类速度快、精度高等特点。
In order to avoid the dependence of the validity of clustering on the space distribution of high dimensional samples of Fuzzy C-Means （ FCM）, a dynamic fuzzy clustering method based on artificial fish swarm algorithm was proposed. By introducing a fuzzy equivalence matrix to the similar degree among samples, the high dimensional samples were mapped to two dimensional planes. Then the Euclidean distance of the samples was approximated to the fuzzy equivalence matrix gradually by using artificial fish swarm algorithm to optimize the coordinate values. Finally, the fuzzy clustering was obtained. The proposed method, not only avoided the dependence of the validity of clustering on the space distribution of high dimensional samples, but also raised the clustering efficiency. Experiment results show that it is an efficient clustering algorithm with rapid speed and high precision.
出处
《计算机应用》
CSCD
北大核心
2009年第6期1569-1571,共3页
journal of Computer Applications
基金
国家民委科研项目(08GX01)
广西自然科学基金资助项目(0832082)
广西民族大学创新计划项目(gxun-chx0885)
关键词
动态模糊聚类
人工鱼群算法
模糊相似矩阵
高维样本
模糊等价矩阵
dynamic fuzzy clustering
artificial fish swarm algorithm
fuzzy similarity matrix
high dimension sample
fuzzy equivalence matrix
