Kernel Self-optimized Locality Preserving Discriminant Analysis for feature extraction and recognition |
| |
Authors: | Jun-Bao LiAuthor Vitae Jeng-Shyang PanAuthor Vitae |
| |
Affiliation: | a Department of Automatic Test and Control, Harbin Institute of Technology, Harbin, China b Department of Electronics, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan, ROC c Department of Computer Science and Information Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, ROC d Graduate Institute of Educational Measurement and Statistics, National Taichung University of Education, Taichung, Taiwan, ROC |
| |
Abstract: | We propose Kernel Self-optimized Locality Preserving Discriminant Analysis (KSLPDA) for feature extraction and recognition. The procedure of KSLPDA is divided into two stages, i.e., one is to solve the optimal expansion of the data-dependent kernel with the proposed kernel self-optimization method, and the second is to seek the optimal projection matrix for dimensionality reduction. Since the optimal parameters of data-dependent kernel are achieved automatically through solving the constraint optimization equation, based on maximum margin criterion and Fisher criterion in the empirical feature space, KSLPDA works well on feature extraction for classification. The comparative experiments show that KSLPDA outperforms PCA, LDA, LPP, supervised LPP and kernel supervised LPP. |
| |
Keywords: | Locality preserving projection Kernel Self-optimized Locality Preserving Discriminant Analysis Kernel method Feature extraction |
本文献已被 ScienceDirect 等数据库收录! |
|