核函数在划分聚类中的应用与实现

Application and Implementation on Partitioning Clustering Using Kernel Function

聚类是数据挖掘的一种重要方法,核函数是能够将低维不可分的数据映射到高维空间进行线性可分时能够降低数据处理难度的重要手段。介绍了聚类算法和核函数的特点。通过引入基于核函数的相似性测度,对k-平均聚类算法和围绕中心点的划分(PAM)算法在Matlab上做了改进和实现。

Clustering is an important way to data mining, and kernel function is an important means of reducing the difficulty of data processing while lower-dimension, non-linear unclassifiable data are mapped to high-dimensional,linear classifiable space.Clustering algorithms and the characteristics of the kernel function are introduced. By introducing a similarity measure based on kernel function, the k-means clustering algorithm and the Partitioning Around Meroid(PAM) algorithm are inproved and imple mented in Matlab.