Flexible non-greedy discriminant subspace feature extraction

Henghao Zhao; Liyong Fu; Zhigang Gao; Qiaolin Ye; Zhangjing Yang; Xubing Yang

doi:10.1016/j.neunet.2019.04.006

Flexible non-greedy discriminant subspace feature extraction

Neural Netw. 2019 Aug:116:166-177. doi: 10.1016/j.neunet.2019.04.006. Epub 2019 Apr 16.

Authors

Henghao Zhao¹, Liyong Fu², Zhigang Gao³, Qiaolin Ye⁴, Zhangjing Yang⁵, Xubing Yang⁶

Affiliations

¹ College of Information Science and Technology, Nanjing Forestry University, Nanjing, Jiangsu 210037, PR China; Key Laboratory of Forest Management and Growth Modeling, National Forestry and Grassland Administration, Beijing 100091, PR China.
² Key Laboratory of Forest Management and Growth Modeling, National Forestry and Grassland Administration, Beijing 100091, PR China; Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing 100091, PR China.
³ College of Animal Science and Technology, Inner Mongolia University for Nationalities, Tongliao, Inner Mongolia 028000, PR China.
⁴ College of Information Science and Technology, Nanjing Forestry University, Nanjing, Jiangsu 210037, PR China. Electronic address: yqlcom@njfu.edu.cn.
⁵ College of Information Engineering, Nanjing Audit University, Nanjing 211815, PR China.
⁶ College of Information Science and Technology, Nanjing Forestry University, Nanjing, Jiangsu 210037, PR China.

PMID: 31063926
DOI: 10.1016/j.neunet.2019.04.006

Abstract

Recently, L₁-norm-based non-greedy linear discriminant analysis (NLDA-L₁) for feature extraction has been shown to be effective for dimensionality reduction, which obtains projection vectors by a non-greedy algorithm. However, it usually acquires unsatisfactory performances due to the utilization of L₁-norm distance measurement. Therefore, in this brief paper, we propose a flexible non-greedy discriminant subspace feature extraction method, which is an extension of NLDA-L₁ by maximizing the ratio of L_p-norm inter-class dispersion to intra-class dispersion. Besides, we put forward a powerful iterative algorithm to solve the resulted objective function and also conduct theoretical analysis on the algorithm. Finally, experimental results on image databases show the effectiveness of our method.

Keywords: Intra-class dispersion; L-norm inter-class dispersion; L-norm-based non-greedy discriminant analysis; Robust distance measurement.

MeSH terms

Algorithms*
Databases, Factual / standards
Discriminant Analysis
Goals
Pattern Recognition, Automated / methods*
Pattern Recognition, Automated / standards