【摘要】 基于构造性核覆盖学习方法的思想,提出了一种构造性核覆盖聚类算法。首先将原空间的待分类样本映射到一个高维的特征空间中,使得样本变得线性可分,然后在核空间采用构造性覆盖方法进行覆盖领域的构造,这组领域能将相似度小的样本分割开来,将相似度大的样本聚合在一起,通过定义一定的相似度度量标准和目标函数,达到聚类的效果。仿真实验也验证了该方法的有效性和可行性。更多还原

【Abstract】 The idea based on the constructive kernel covering study means,puts forward a constructive kernel covering clustering algorithm.Firstly,it maps the awaiting sort samples of original space to a high dimensional feature space,makes the samples linear separable.Then,in kernel space,the algorithm uses constructive covering method to construct covering domain.Every domain divides the minor semblance samples,makes the major semblance samples converge.By defining definite semblance standard and target function,it can get the clustering effect.Emulation experiment proved the validity and the feasibility of this algorithm.更多还原