Faster Principal Component Regression and Stable Matrix Chebyshev Approximation

Abstract

We solve principal component regression (PCR), up to a multiplicative accuracy $1+\gamma$, by reducing the problem to $\tilde{O}(\gamma^{-1})$ black-box calls of ridge regression. Therefore, our algorithm does not require any explicit construction of the top principal components, and is suitable for large-scale PCR instances. In contrast, previous result requires $\tilde{O}(\gamma^{-2})$ such black-box calls. We obtain this result by developing a general stable recurrence formula for matrix Chebyshev polynomials, and a degree-optimal polynomial approximation to the matrix sign function. Our techniques may be of independent interests, especially when designing iterative methods.

Cite

Text

Allen-Zhu and Li. "Faster Principal Component Regression and Stable Matrix Chebyshev Approximation." International Conference on Machine Learning, 2017.

Markdown

[Allen-Zhu and Li. "Faster Principal Component Regression and Stable Matrix Chebyshev Approximation." International Conference on Machine Learning, 2017.](https://mlanthology.org/icml/2017/allenzhu2017icml-faster/)

BibTeX

@inproceedings{allenzhu2017icml-faster,
  title     = {{Faster Principal Component Regression and Stable Matrix Chebyshev Approximation}},
  author    = {Allen-Zhu, Zeyuan and Li, Yuanzhi},
  booktitle = {International Conference on Machine Learning},
  year      = {2017},
  pages     = {107-115},
  volume    = {70},
  url       = {https://mlanthology.org/icml/2017/allenzhu2017icml-faster/}
}