Randomly Aggregated Least Squares for Support Recovery

Abstract
We study the problem of exact support recovery: given an (unknown) vector , we are given access to the noisy measurement y = X\theta + \omega, where is a (known) Gaussian matrix and the noise is an (unknown) Gaussian vector. How small we can choose and still reliably recover the support of ? We present RAWLS (Randomly Aggregated UnWeighted Least Squares Support Recovery): the main idea is to take random subsets of the equations, perform a least squares recovery over this reduced bit of information and then average over many random subsets. We show that the proposed procedure can provably recover an approximation of and demonstrate its use in support recovery through numerical examples.
View on arXivComments on this paper