有监督的核局部线性嵌入算法  

Supervised Kernel Locally Linear Embedding

作　　者：喻军[1] 申家峰[1] 

机构地区：[1]郑州航空工业管理学院数理系,郑州450015

出　　处：《科学技术与工程》2008年第14期3982-3984,共3页Science Technology and Engineering

基　　金：郑州航空工业管理学院青年基金(Q07L068)资助

摘　　要：核局部线性嵌入法是一个优异的流形学习方法,对于非线性高维数据的降维问题,具有较好的效果。但是算法本身是一个无监督学习方法,对于模式分类等有监督学习问题效果不是很好。通过分析监督学习问题的机理,提出了一种有监督的核函数局部线性嵌入算法,数值实验证明算法对于有监督学习问题,具有较好的效果。Kernel locally linear embedding is an excellent manifold learning methods. It perform well for non- linear data dimensionality reduction. But the methods is an unsupervised learning algorithms, its effect is not good for supervised learning problems, such as pattern recognition. By analyzing the mechanism of supervised learning, a supervised kernel locally linear embedding method is proposd. Data experiment show that the algorithm performs well for supervised learning problems.

关 键 词：核局部线性嵌入 流形学习 有监督学习 

分 类 号：TP368.1[自动化与计算机技术—计算机系统结构]