In this paper, we propose and analyze zeroth-order stochastic approximation algorithms for nonconvex and convex optimization. Specifically, we propose generalizations of the conditional gradient algorithm achieving rates similar to the standard stochastic gradient algorithm using only zeroth-order information. Furthermore, under a structural sparsity assumption, we first illustrate an implicit regularization phenomenon where the standard stochastic gradient algorithm with zeroth-order information adapts to the sparsity of the problem at hand by just varying the step-size. Next, we propose a truncated stochastic gradient algorithm with zeroth-order information, whose rate of convergence depends only poly-logarithmically on the dimensionality.
Krishnakumar Balasubramanian (University of California, Davis)
Saeed Ghadimi (Princeton University)
More from the Same Authors
2017 Poster: Estimating High-dimensional Non-Gaussian Multiple Index Models via Stein’s Lemma »
Zhuoran Yang · Krishnakumar Balasubramanian · Zhaoran Wang · Han Liu