Optimal prediction for sparse linear models? Lower bounds for coordinate-separable M-estimators
Abstract
For the problem of high-dimensional sparse linear regression, it is known that an -based estimator can achieve a "fast" rate on the prediction error without any conditions on the design matrix, whereas in absence of restrictive conditions on the design matrix, popular polynomial-time methods only guarantee the "slow" rate. In this paper, we show that the slow rate is intrinsic to a broad class of M-estimators. In particular, for estimators based on minimizing a least-squares cost function together with a (possibly non-convex) coordinate-wise separable regularizer, there is always a "bad" local optimum such that the associated prediction error is lower bounded by a constant multiple of . For convex regularizers, this lower bound applies to all global optima. The theory is applicable to many popular estimators, including convex -based methods as well as M-estimators based on nonconvex regularizers, including the SCAD penalty or the MCP regularizer. In addition, for a broad class of nonconvex regularizers, we show that the bad local optima are very common, in that a broad class of local minimization algorithms with random initialization will typically converge to a bad solution.
Cite
@article{arxiv.1503.03188,
title = {Optimal prediction for sparse linear models? Lower bounds for coordinate-separable M-estimators},
author = {Yuchen Zhang and Martin J. Wainwright and Michael I. Jordan},
journal= {arXiv preprint arXiv:1503.03188},
year = {2015}
}
Comments
Add more coverage on related work; add a new lower bound for design matrices satisfying the restricted eigenvalue condition