Near-Optimal Algorithms for Convex Simple Bilevel Optimization under Weak Assumptions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jiang, Rujun, Shi, Xu, Song, Weizheng, Wang, Jiulin
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913935445196800
author Jiang, Rujun
Shi, Xu
Song, Weizheng
Wang, Jiulin
author_facet Jiang, Rujun
Shi, Xu
Song, Weizheng
Wang, Jiulin
contents This paper considers the simple bilevel optimization (SBO) problem, which minimizes a composite convex function over the optimal solution set of another composite convex minimization problem. We first show that this bilevel problem is equivalent to finding the left-most root of a nonlinear equation. Based on this and a novel dual approach for solving the subproblem in each iteration, we efficiently obtain an $(ε, ε)$-optimal solution through the bisection and Newton methods. The proposed methods achieve near-optimal operation complexity of ${\tilde{\mathcal{O}}(\sqrt{1/ε})}$ under mild assumptions, aligning with the lower complexity bounds of the first-order methods in SBO with both level objectives being smooth convex and unconstrained composite convex optimization when ignoring logarithmic terms.
format Preprint
id arxiv_https___arxiv_org_abs_2409_08948
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Near-Optimal Algorithms for Convex Simple Bilevel Optimization under Weak Assumptions
Jiang, Rujun
Shi, Xu
Song, Weizheng
Wang, Jiulin
Optimization and Control
This paper considers the simple bilevel optimization (SBO) problem, which minimizes a composite convex function over the optimal solution set of another composite convex minimization problem. We first show that this bilevel problem is equivalent to finding the left-most root of a nonlinear equation. Based on this and a novel dual approach for solving the subproblem in each iteration, we efficiently obtain an $(ε, ε)$-optimal solution through the bisection and Newton methods. The proposed methods achieve near-optimal operation complexity of ${\tilde{\mathcal{O}}(\sqrt{1/ε})}$ under mild assumptions, aligning with the lower complexity bounds of the first-order methods in SBO with both level objectives being smooth convex and unconstrained composite convex optimization when ignoring logarithmic terms.
title Near-Optimal Algorithms for Convex Simple Bilevel Optimization under Weak Assumptions
topic Optimization and Control
url https://arxiv.org/abs/2409.08948