摘要
为了对含有噪声和离群点的多特征类样本数据进行有效的聚类,提出了一种基于多核的改进模糊聚类算法。该算法选取子核函数构造多核函数,将输入的样本经多核函数进行映射,增加了不同类别样本间的区分度,提高核函数的学习能力和泛化能力。实验结果表明,该算法对于多样本数据具有比单核更好的聚类效果。
For the effective clustering of multi-feature sample data that contain noise and outliers, an improved fuzzy clustering al-gorithm based on multi-core is proposed. The algorithm selected sub kernel function to construct multi-core function, and mappedthe input samples by multi-core function, which increases the distinguish of different categories of samples, and improves the learn-ing ability and generalization ability of kernel function. The experimental results have show that the algorithm has a better cluster-ing effect than single core for multi sample data.
出处
《电脑知识与技术》
2018年第5X期7-9,共3页
Computer Knowledge and Technology
基金
河南省重点科技攻关项目(142102210231)
校级项目(GKJ2017016)
关键词
聚类
核函数
多核函数
multi-view
clustering
spectral clustering