关键词: 多核, 模糊核聚类, 高斯核, Sigmoid核, 多项式核函数

Abstract: Because the fuzzy clustering based on single kernel function has some limitation on performance, a new multiple-kernel fuzzy clustering is put forward. The new multiple kernel is constituted by Gaussian kernel, Sigmoid kernel and polynomial kernel. Gaussian kernel is wildly used in KFCM. It can be demonstrated that sigmoid kernel derived from neural network has good global classification performance, as well as polynomial kernel. The new multiple kernel combines the advantage of them. The experimental results prove that the fuzzy clustering with multiple kernel is much better than with single kernel on performance.

Key words: multiple kernel, Kernel Fuzzy Clustering（KFCM）, Gaussian kernel, Sigmoid kernel, polynomial kernel