监督型局部保持的典型相关分析

Supervised Locality Preserving Canonical Correlation Analysis

利用数据集的局部结构信息和判别结构信息,构建相似度矩阵和类信息矩阵,提出监督型局部保持的典型相关分析(Supervised Locality Preserving Canonical Correlation Analysis,SLPCCA),该方法不但突破了典型相关分析(Canonical Correla-tion Analysis,CCA)处理数据时的线性约束,提高了处理非线性问题的能力,而且克服了局部保持的典型相关分析(LocalityPreserving Canonical Correlation Analysis,LPCCA)忽视类信息的问题,提取的特征更有利于分类.在多特征手写体数据库(MFD)和美国国家邮政局手写字库(USPS)上的实验结果验证了该算法的有效性.

In this paper,a novel supervised locality preserving canonical correlation analysis（SLPCCA）is developed,which uses discriminative structural information to construct the class-information matrix,as well as combine the correlation of the neighboring samples to construct the similarity matrix.As a result,SLPCCA can not only improve the ability of CCA to solve nonlinear problems by infusing the local structural information and breaking the linear restriction,but also overcome the shortcoming of LPCCA which neglects the class information.We obtain features which is more favor of classification compared with LPCCA.The experimental results on Multiple Feature Dataset and USPS Dataset have demonstrated the superiority of our proposed SLPCCA compared with CCA and LPCCA.