基于高斯隶属度的包容性指标模糊聚类算法
The fuzzy clustering algorithm based on inclusion index of Gausian membership function

摘要： 基于传统的模糊聚类算法(C-means、FCM),在高斯隶属度函数的基础上给出了包含性指标的定义,提出了基于高斯隶属度的包容性指标模糊聚类算法(fuzzy inclusion-based clustering, FIC)。该方法通过获取高斯隶属度函数的包含性指标,为每个分类确定一个支持距离的半定性矩阵,来保证每个分类到所有数据类的距离和与所有数据类包含度的总和一致。通过UCI中Wine数据集进行了仿真实验,实验结果表明与FCM算法相比较,FIC算法具有更好的有效性和可行性。
Abstract: Based on traditional fuzzy clustering algorithm, such as Fuzzy algorithm and C-means algorithm, the definition of inclusion index is taken into account the fuzzy clustering algorithm proposed, which is based on Gaussian membership functions. This algorithm ensure that the distance of each classification to another classifications is the same by a semi definite matrix, which preserves the inclusion index of Gausian membership function. The simulation experiment results with the Wine data set of UCI show that, compared with FCM, FIC algorithm has more effectiveness and feasibility