Non-Gaussian Component Analysis: A Semi-Parametric Framework for Linear Dimension Reduction

Abstract

We propose a new linear method for dimension reduction to identify nonGaussian components in high dimensional data. Our method, NGCA (non-Gaussian component analysis), uses a very general semi-parametric framework. In contrast to existing projection methods we define what is uninteresting (Gaussian): by projecting out uninterestingness, we can estimate the relevant non-Gaussian subspace. We show that the estimation error of finding the non-Gaussian components tends to zero at a parametric rate. Once NGCA components are identified and extracted, various tasks can be applied in the data analysis process, like data visualization, clustering, denoising or classification. A numerical study demonstrates the usefulness of our method.

Cite

Text

Blanchard et al. "Non-Gaussian Component Analysis: A Semi-Parametric Framework for Linear Dimension Reduction." Neural Information Processing Systems, 2005.

Markdown

[Blanchard et al. "Non-Gaussian Component Analysis: A Semi-Parametric Framework for Linear Dimension Reduction." Neural Information Processing Systems, 2005.](https://mlanthology.org/neurips/2005/blanchard2005neurips-nongaussian/)

BibTeX

@inproceedings{blanchard2005neurips-nongaussian,
  title     = {{Non-Gaussian Component Analysis: A Semi-Parametric Framework for Linear Dimension Reduction}},
  author    = {Blanchard, Gilles and Sugiyama, Masashi and Kawanabe, Motoaki and Spokoiny, Vladimir and Müller, Klaus-Robert},
  booktitle = {Neural Information Processing Systems},
  year      = {2005},
  pages     = {131-138},
  url       = {https://mlanthology.org/neurips/2005/blanchard2005neurips-nongaussian/}
}