211ab571cc9f3802afa6ffff52ae3e5b-Paper-Conference.pdf

In addition, the underlying signalxisassumed to lie in the range of anL-Lipschitz continuous generativemodel with boundedkdimensionalinputs.Weproposeatwo-stepapproach,forwhichthefirststepplays the role ofspectral initialization and the second step refines the estimated vector produced by the first step iteratively. We show that both steps enjoy a statistical rate oforder p (klogL) (logm)/mundersuitable conditions.