Kernel Feature Selection via Conditional Covariance Minimization

Abstract

We propose a method for feature selection that employs kernel-based measures of independence to find a subset of covariates that is maximally predictive of the response. Building on past work in kernel dimension reduction, we show how to perform feature selection via a constrained optimization problem involving the trace of the conditional covariance operator. We prove various consistency results for this procedure, and also demonstrate that our method compares favorably with other state-of-the-art algorithms on a variety of synthetic and real data sets.

Cite

Text

Chen et al. "Kernel Feature Selection via Conditional Covariance Minimization." Neural Information Processing Systems, 2017.

Markdown

[Chen et al. "Kernel Feature Selection via Conditional Covariance Minimization." Neural Information Processing Systems, 2017.](https://mlanthology.org/neurips/2017/chen2017neurips-kernel/)

BibTeX

@inproceedings{chen2017neurips-kernel,
  title     = {{Kernel Feature Selection via Conditional Covariance Minimization}},
  author    = {Chen, Jianbo and Stern, Mitchell and Wainwright, Martin J. and Jordan, Michael I},
  booktitle = {Neural Information Processing Systems},
  year      = {2017},
  pages     = {6946-6955},
  url       = {https://mlanthology.org/neurips/2017/chen2017neurips-kernel/}
}