基于局部与全局保持的半监督维数约减方法

Local and Global Preserving Based Semi-Supervised Dimensionality Reduction Method

在很多机器学习和数据挖掘任务中,仅仅利用边信息(side-information)并不能得到最好的半监督学习(semi-supervised learning)效果,因此,提出一种基于局部与全局保持的半监督维数约减(local and global preserving based semi-supervised dimensionality reduction,简称LGSSDR)方法.该算法不仅能够保持正、负约束信息而且能够保持数据集所在低维流形的全局以及局部信息.另外,该算法能够计算出变换矩阵并较容易地处理未见样本.实验结果验证了该算法的有效性.

In many machine learning and data mining tasks, it can＇t achieve the best semi-supervised learning result if only use side-information. So, a local and global preserving based semi-supervised dimensionality reduction （LGSSDR） method is proposed in this paper. LGSSDR algorithm can not only preserve the positive and negative constraints but also preserve the local and global structure of the whole data manifold in the low dimensional embedding subspace. Besides, the algorithm can compute the transformation matrix and handle unseen samples easily. Experimental results on several datasets demonstrate the effectiveness of this method.