核K-均值聚类(Kernel K-means Clustering)

时间:2022-01-06 22:52:25

问题:

       设数据集核K-均值聚类(Kernel K-means Clustering),其中核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering)。Mercer核函数核K-均值聚类(Kernel K-means Clustering),根据Mercer定理存在映射核K-均值聚类(Kernel K-means Clustering),使得核K-均值聚类(Kernel K-means Clustering)

        核K-均值聚类就是讨论映射数据集核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering)空间中的聚类情况,设在核K-均值聚类(Kernel K-means Clustering)空间中,把数据集分为核K-均值聚类(Kernel K-means Clustering)类,核K-均值聚类(Kernel K-means Clustering)为第核K-均值聚类(Kernel K-means Clustering)类的均值,核K-均值聚类(Kernel K-means Clustering)

即考虑以下模型:

核K-均值聚类(Kernel K-means Clustering)

核K-均值聚类(Kernel K-means Clustering)


问题1:

怎么训练上述模型,因为核K-均值聚类(Kernel K-means Clustering)一般情况下是解不出来的。

方法:

初始化核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering),其中核K-均值聚类(Kernel K-means Clustering)

核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering)

E步:求核K-均值聚类(Kernel K-means Clustering)

核K-均值聚类(Kernel K-means Clustering)

注意其中:

核K-均值聚类(Kernel K-means Clustering)核K-均值聚类(Kernel K-means Clustering)

M步:固定核K-均值聚类(Kernel K-means Clustering),求核K-均值聚类(Kernel K-means Clustering)

核K-均值聚类(Kernel K-means Clustering)

核K-均值聚类(Kernel K-means Clustering),

核K-均值聚类(Kernel K-means Clustering)

其中核K-均值聚类(Kernel K-means Clustering)

进入下一轮迭代,直至收敛!