English

Principal Component Analysis Based on T$\ell_1$-norm Maximization

Machine Learning 2020-05-27 v1 Machine Learning

Abstract

Classical principal component analysis (PCA) may suffer from the sensitivity to outliers and noise. Therefore PCA based on 1\ell_1-norm and p\ell_p-norm (0<p<10 < p < 1) have been studied. Among them, the ones based on p\ell_p-norm seem to be most interesting from the robustness point of view. However, their numerical performance is not satisfactory. Note that, although T1\ell_1-norm is similar to p\ell_p-norm (0<p<10 < p < 1) in some sense, it has the stronger suppression effect to outliers and better continuity. So PCA based on T1\ell_1-norm is proposed in this paper. Our numerical experiments have shown that its performance is superior than PCA-p\ell_p and p\ell_pSPCA as well as PCA, PCA-1\ell_1 obviously.

Keywords

Cite

@article{arxiv.2005.12263,
  title  = {Principal Component Analysis Based on T$\ell_1$-norm Maximization},
  author = {Xiang-Fei Yang and Yuan-Hai Shao and Chun-Na Li and Li-Ming Liu and Nai-Yang Deng},
  journal= {arXiv preprint arXiv:2005.12263},
  year   = {2020}
}
R2 v1 2026-06-23T15:47:53.363Z