摘要
由于现有模糊C-均值聚类算法固有的局限性,本文提出了一种改进的模糊C-均值聚类算法.首先用概率密度函数来确定初始聚类中心点和聚类数,其次用竞争学习思想提出使对手增加抑制因子来修改隶属度得到加快收敛速度的效果,最后提出用一个类内差异与类间差异兼备的新的有效性指标来作为迭代条件的目标函数.通过实验获取参数的最优取值范围,通过与经典模糊C-均值聚类算法的比较,证明了该改进算法不仅加快了收敛速度,而且在聚类结果的质量上有一定程度的提高.
The fuzzy C-means algorithm was improved to break through the existing performance limitations. The function of probability consistency was used to determine the original clustering center and the clustering number. For each object an inhibitory factor was added to its opponent to accelerate the convergence. A new validity index which takes a balance between intra-clustering and inter-clustering variation was proposed to act as the aim function. Experiments show that the improved algorithm behaves comparatively higher performance in convergence speed and clustering quality.
出处
《上海理工大学学报》
CAS
北大核心
2012年第4期351-354,共4页
Journal of University of Shanghai For Science and Technology
基金
国家自然科学基金资助项目(70971089)
上海市重点学科建设资助项目(S30501)
关键词
模糊C-均值聚类
概率密度
隶属度
有效性指标
fuzzy C-means clustering； probability consistency； subjection value； validity index；
