Saved in:
Bibliographic Details
Main Authors: Gupta, Shuvomoy Das, Stellato, Bartolomeo, Van Parys, Bart P. G.
Format: Preprint
Published: 2020
Subjects:
Online Access:https://arxiv.org/abs/2011.04552
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911855345139712
author Gupta, Shuvomoy Das
Stellato, Bartolomeo
Van Parys, Bart P. G.
author_facet Gupta, Shuvomoy Das
Stellato, Bartolomeo
Van Parys, Bart P. G.
contents Many problems of substantial current interest in machine learning, statistics, and data science can be formulated as sparse and low-rank optimization problems. In this paper, we present the nonconvex exterior-point optimization solver NExOS -- a first-order algorithm tailored to sparse and low-rank optimization problems. We consider the problem of minimizing a convex function over a nonconvex constraint set, where the set can be decomposed as the intersection of a compact convex set and a nonconvex set involving sparse or low-rank constraints. Unlike the convex relaxation approaches, NExOS finds a locally optimal point of the original problem by solving a sequence of penalized problems with strictly decreasing penalty parameters by exploiting the nonconvex geometry. NExOS solves each penalized problem by applying a first-order algorithm, which converges linearly to a local minimum of the corresponding penalized formulation under regularity conditions. Furthermore, the local minima of the penalized problems converge to a local minimum of the original problem as the penalty parameter goes to zero. We then implement and test NExOS on many instances from a wide variety of sparse and low-rank optimization problems, empirically demonstrating that our algorithm outperforms specialized methods.
format Preprint
id arxiv_https___arxiv_org_abs_2011_04552
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Exterior-point Optimization for Sparse and Low-rank Optimization
Gupta, Shuvomoy Das
Stellato, Bartolomeo
Van Parys, Bart P. G.
Optimization and Control
Many problems of substantial current interest in machine learning, statistics, and data science can be formulated as sparse and low-rank optimization problems. In this paper, we present the nonconvex exterior-point optimization solver NExOS -- a first-order algorithm tailored to sparse and low-rank optimization problems. We consider the problem of minimizing a convex function over a nonconvex constraint set, where the set can be decomposed as the intersection of a compact convex set and a nonconvex set involving sparse or low-rank constraints. Unlike the convex relaxation approaches, NExOS finds a locally optimal point of the original problem by solving a sequence of penalized problems with strictly decreasing penalty parameters by exploiting the nonconvex geometry. NExOS solves each penalized problem by applying a first-order algorithm, which converges linearly to a local minimum of the corresponding penalized formulation under regularity conditions. Furthermore, the local minima of the penalized problems converge to a local minimum of the original problem as the penalty parameter goes to zero. We then implement and test NExOS on many instances from a wide variety of sparse and low-rank optimization problems, empirically demonstrating that our algorithm outperforms specialized methods.
title Exterior-point Optimization for Sparse and Low-rank Optimization
topic Optimization and Control
url https://arxiv.org/abs/2011.04552