摘要
流形学习方法是一种新型的非线性降维方法,它可以有效地对具有内在流形形式的非线性高维数据进行维数约简.目前,流形学习已被成功应用于聚类、可视化等数据挖掘领域,表现出卓越的性能.首先讨论了流形学习的研究现状,然后介绍了这一领域中影响最大的2种算法:局部线性嵌入算法和等距特征映射算法.
The manifold learning method is a new kind of nonlinear dimensionality reduction approach, which can effectively reduce dimensions for high dimensional data in an intrinsic nonlinear manifold form. Until currently, this kind of method has been successfully applied to many data mining areas such as clustering and visualization and exhibit powerful capability in these applications. This article concentrates the research on the theory of manifold learning, and discusses two algorithms, Isometric Mapping algorithm and Locally Linear Embedding algorithm, which are two of the most effective manifold learning methods.
出处
《云南民族大学学报（自然科学版）》
CAS
2008年第4期370-373,共4页
Journal of Yunnan Minzu University:Natural Sciences Edition
基金
云南省自然科学基金资助项目(2005F0028Q)
国家民族事务委员会科研资助项目(08YN02)
云南省教育厅科研基金资助项目(6Y0006D)
