Collaborative representation analysis methods for feature extraction

Juliang Hua, Huan Wang, Mingu Ren, Heyan Huang

Abstract

Recently, sparse representation (SR) theory gets much success in the fields of pattern recognition and machine learning. Many researchers use SR to design classification methods and dictionary learning via reconstruction residual. It was shown that collaborative representation (CR) is the key part in sparse representation-based classification (SRC) and collaborative representation-based classification (CRC). Both SRC and CRC are good classification methods. Here, we give a collaborative representation analysis (CRA) method for feature extraction. Not like SRC-/CRC-based methods (e.g., SPP and CRP), CRA could directly extract the features like PCA and LDA. Further, a Kernel CRA (KCRA) is developed via kernel tricks. The experimental results on FERET and AR face databases show that CRA and KCRA are two effective feature extraction methods and could get good performance.
BibTeX
@article{hua2017collaborative,
  author = {Juliang Hua and Huan Wang and Mingu Ren and Heyan Huang},
  title = {{Collaborative representation analysis methods for feature extraction}},
  journal = {Neural Computing and Applications},
  year = {2017},
  doi = {10.1007/s00521-016-2299-3},
}

Loading related content...

Figure