Minimax Rates of Estimation for Sparse PCA in High Dimensions

Abstract

We study sparse principal components analysis in the high-dimensional setting, where p (the number of variables) can be much larger than n (the number of observations). We prove optimal minimax lower and upper bounds on the estimation error for the first leading eigenvector when it belongs to an \ell_q ball for q ∈[0,1]. Our bounds are tight in p and n for all q ∈[0, 1] over a wide class of distributions. The upper bound is obtained by analyzing the performance of \ell_q-constrained PCA. In particular, our results provide convergence rates for \ell_1-constrained PCA. Our methods and arguments are also extendable to multi-dimensional subspace estimation.

Cite

Text

Vu and Lei. "Minimax Rates of Estimation for Sparse PCA in High Dimensions." Proceedings of the Fifteenth International Conference on Artificial Intelligence and Statistics, 2012.

Markdown

[Vu and Lei. "Minimax Rates of Estimation for Sparse PCA in High Dimensions." Proceedings of the Fifteenth International Conference on Artificial Intelligence and Statistics, 2012.](https://mlanthology.org/aistats/2012/vu2012aistats-minimax/)

BibTeX

@inproceedings{vu2012aistats-minimax,
  title     = {{Minimax Rates of Estimation for Sparse PCA in High Dimensions}},
  author    = {Vu, Vincent and Lei, Jing},
  booktitle = {Proceedings of the Fifteenth International Conference on Artificial Intelligence and Statistics},
  year      = {2012},
  pages     = {1278-1286},
  volume    = {22},
  url       = {https://mlanthology.org/aistats/2012/vu2012aistats-minimax/}
}