skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: A Conditional Gradient-based Method for Simple Bilevel Optimization with Convex Lower-level Problem
In this paper, we study a class of bilevel optimization problems, also known as simple bilevel optimization, where we minimize a smooth objective function over the optimal solution set of another convex constrained optimization problem. Several iterative methods have been developed for tackling this class of problems. Alas, their convergence guarantees are either asymptotic for the upper-level objective, or the convergence rates are slow and sub-optimal. To address this issue, in this paper, we introduce a novel bilevel optimization method that locally approximates the solution set of the lower-level problem via a cutting plane and then runs a conditional gradient update to decrease the upper-level objective. When the upper-level objective is convex, we show that our method requires $${O}(\max\{1/\epsilon_f,1/\epsilon_g\})$$ iterations to find a solution that is $$\epsilon_f$$-optimal for the upper-level objective and $$\epsilon_g$$-optimal for the lower-level objective. Moreover, when the upper-level objective is non-convex, our method requires $${O}(\max\{1/\epsilon_f^2,1/(\epsilon_f\epsilon_g)\})$$ iterations to find an $$(\epsilon_f,\epsilon_g)$$-optimal solution. We also prove stronger convergence guarantees under the Holderian error bound assumption on the lower-level problem. To the best of our knowledge, our method achieves the best-known iteration complexity for the considered class of bilevel problems.  more » « less
Award ID(s):
2127696
PAR ID:
10443637
Author(s) / Creator(s):
; ; ;
Editor(s):
Ruiz, Francisco; Dy, Jennifer; van de Meent, Jan-Willem
Date Published:
Journal Name:
Proceedings of Machine Learning Research
Volume:
206
ISSN:
2640-3498
Page Range / eLocation ID:
10305 - 10323
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. (, Advances in neural information processing systems)
    This work proposes a new algorithm – the Single-timescale Double-momentum Stochastic Approximation (SUSTAIN) –for tackling stochastic unconstrained bilevel optimization problems. We focus on bilevel problems where the lower level subproblem is strongly-convex and the upper level objective function is smooth. Unlike prior works which rely on two-timescale or double loop techniques, we design a stochastic momentum-assisted gradient estimator for both the upper and lower level updates. The latter allows us to control the error in the stochastic gradient updates due to inaccurate solution to both subproblems. If the upper objective function is smooth but possibly non-convex, we show that SUSTAIN requires $${O}(\epsilon^{-3/2})$$ iterations (each using $O(1)$ samples) to find an $$\epsilon$$-stationary solution. The $$\epsilon$$-stationary solution is defined as the point whose squared norm of the gradient of the outer function is less than or equal to $$\epsilon$$. The total number of stochastic gradient samples required for the upper and lower level objective functions match the best-known complexity for single-level stochastic gradient algorithms. We also analyze the case when the upper level objective function is strongly-convex. 
    more » « less
  2. ; ; (, Advances in Neural Information Processing Systems)
    In this paper, we revisit the bilevel optimization problem, in which the upper-level objective function is generally nonconvex and the lower-level objective function is strongly convex. Although this type of problem has been studied extensively, it still remains an open question how to achieve an $$\mathcal{O}(\epsilon^{-1.5})$$ sample complexity in Hessian/Jacobian-free stochastic bilevel optimization without any second-order derivative computation. To fill this gap, we propose a novel Hessian/Jacobian-free bilevel optimizer named FdeHBO, which features a simple fully single-loop structure, a projection-aided finite-difference Hessian/Jacobian-vector approximation, and momentum-based updates. Theoretically, we show that FdeHBO requires $$\mathcal{O}(\epsilon^{-1.5})$$ iterations (each using $$\mathcal{O}(1)$$ samples and only first-order gradient information) to find an $$\epsilon$$-accurate stationary point. As far as we know, this is the first Hessian/Jacobian-free method with an $$\mathcal{O}(\epsilon^{-1.5})$$ sample complexity for nonconvex-strongly-convex stochastic bilevel optimization. 
    more » « less
  3. ; (, Proceedings of Machine Learning Research 157, 2021)
    Solving a bilevel optimization problem is at the core of several machine learning problems such as hyperparameter tuning, data denoising, meta- and few-shot learning, and training-data poisoning. Different from simultaneous or multi-objective optimization, the steepest descent direction for minimizing the upper-level cost in a bilevel problem requires the inverse of the Hessian of the lower-level cost. In this work, we propose a novel algorithm for solving bilevel optimization problems based on the classical penalty function approach. Our method avoids computing the Hessian inverse and can handle constrained bilevel problems easily. We prove the convergence of the method under mild conditions and show that the exact hypergradient is obtained asymptotically. Our method's simplicity and small space and time complexities enable us to effectively solve large-scale bilevel problems involving deep neural networks. We present results on data denoising, few-shot learning, and training-data poisoning problems in a large-scale setting. Our results show that our approach outperforms or is comparable to previously proposed methods based on automatic differentiation and approximate inversion in terms of accuracy, run-time, and convergence speed. 
    more » « less
  4. ; (, Operations Research)
    In a chance constrained program (CCP), decision makers seek the best decision whose probability of violating the uncertainty constraints is within the prespecified risk level. As a CCP is often nonconvex and is difficult to solve to optimality, much effort has been devoted to developing convex inner approximations for a CCP, among which the conditional value-at-risk (CVaR) has been known to be the best for more than a decade. This paper studies and generalizes the ALSO-X, originally proposed by Ahmed, Luedtke, SOng, and Xie in 2017 , for solving a CCP. We first show that the ALSO-X resembles a bilevel optimization, where the upper-level problem is to find the best objective function value and enforce the feasibility of a CCP for a given decision from the lower-level problem, and the lower-level problem is to minimize the expectation of constraint violations subject to the upper bound of the objective function value provided by the upper-level problem. This interpretation motivates us to prove that when uncertain constraints are convex in the decision variables, ALSO-X always outperforms the CVaR approximation. We further show (i) sufficient conditions under which ALSO-X can recover an optimal solution to a CCP; (ii) an equivalent bilinear programming formulation of a CCP, inspiring us to enhance ALSO-X with a convergent alternating minimization method (ALSO-X+); and (iii) an extension of ALSO-X and ALSO-X+ to distributionally robust chance constrained programs (DRCCPs) under the ∞−Wasserstein ambiguity set. Our numerical study demonstrates the effectiveness of the proposed methods. 
    more » « less
  5. ; ; (, Decision Analysis)
    null (Ed.)
    Traditionally, in the bilevel optimization framework, a leader chooses her actions by solving an upper-level problem, assuming that a follower chooses an optimal reaction by solving a lower-level problem. However, in many settings, the lower-level problems might be nontrivial, thus requiring the use of tailored algorithms for their solution. More importantly, in practice, such problems might be inexactly solved by heuristics and approximation algorithms. Motivated by this consideration, we study a broad class of bilevel optimization problems where the follower might not optimally react to the leader’s actions. In particular, we present a modeling framework in which the leader considers that the follower might use one of a number of known algorithms to solve the lower-level problem, either approximately or heuristically. Thus, the leader can hedge against the follower’s use of suboptimal solutions. We provide algorithmic implementations of the framework for a class of nonlinear bilevel knapsack problem (BKP), and we illustrate the potential impact of incorporating this realistic feature through numerical experiments in the context of defender-attacker problems. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email