Linearly convergent bilevel optimization with single-step inner methods
From MaRDI portal
Publication:6155063
DOI10.1007/S10589-023-00527-7arXiv2205.04862OpenAlexW4387131582MaRDI QIDQ6155063FDOQ6155063
Authors: Ensio Suonperä, Tuomo Valkonen 
Publication date: 16 February 2024
Published in: Computational Optimization and Applications (Search for Journal in Brave)
Abstract: We propose a new approach to solving bilevel optimization problems, intermediate between solving full-system optimality conditions with a Newton-type approach, and treating the inner problem as an implicit function. The overall idea is to solve the full-system optimality conditions, but to precondition them to alternate between taking steps of simple conventional methods for the inner problem, the adjoint equation, and the outer problem. While the inner objective has to be smooth, the outer objective may be nonsmooth subject to a prox-contractivity condition. We prove linear convergence of the approach for combinations of gradient descent and forward-backward splitting with exact and inexact solution of the adjoint equation. We demonstrate good performance on learning the regularization parameter for anisotropic total variation image denoising, and the convolution kernel for image deconvolution.
Full work available at URL: https://arxiv.org/abs/2205.04862
Cites Work
- Title not available (Why is that?)
- A first-order primal-dual algorithm for convex problems with applications to imaging
- Image denoising: learning the noise model via nonsmooth PDE-constrained optimization
- Bilevel optimization for calibrating point spread functions in blind deconvolution
- A bilevel optimization approach for parameter learning in variational models
- Mathematical Programs with Equilibrium Constraints
- Solving bilevel programs with the KKT-approach
- Convergence analysis of primal-dual algorithms for a saddle-point problem: from contraction perspective
- Bi-CGSTAB: A Fast and Smoothly Converging Variant of Bi-CG for the Solution of Nonsymmetric Linear Systems
- Optimality conditions for bilevel programming problems
- Bilevel programming problems. Theory, algorithms and applications to energy networks
- Application of particle swarm optimization based on CHKS smoothing function for solving nonlinear bilevel programming problem
- The bilevel programming problem: reformulations, constraint qualifications and optimality conditions
- A primal-dual hybrid gradient method for nonlinear operators with applications to MRI
- Bilevel parameter learning for higher-order total variation regularisation models
- Title not available (Why is that?)
- Dynamic sampling schemes for optimal noise learning under multiple nonsmooth constraints
- Solving ill-posed bilevel programs
- Techniques for gradient-based bilevel optimization with non-smooth lower level problems
- The structure of optimal parameters for image restoration problems
- Sufficient optimality conditions in bilevel programming
- On bilevel programming. I: General nonlinear cases
- An inertial extrapolation method for convex simple bilevel optimization
- A first order method for solving convex bilevel optimization problems
- Optimality Conditions for Bilevel Imaging Learning Problems with Total Variation Regularization
- Optimal selection of the regularization function in a weighted total variation model. II: Algorithm, its analysis and numerical tests
- Generalized Nash equilibrium problems, bilevel programming and MPEC. Based on lectures given at the international center for pure and applied mathematics (CIMPA) school, Delhi, India, November 25 -- December 6, 2013
- Gauss-Newton-type methods for bilevel optimization
- Testing and non-linear preconditioning of the proximal point method
- Inexact derivative-free optimization for bilevel learning
- A bilevel approach for parameter learning in inverse problems
- Acceleration and Global Convergence of a First-Order Primal-Dual Method for Nonconvex Problems
- Learning consistent discretizations of the total variation
- Directional necessary optimality conditions for bilevel programs
Cited In (1)
This page was built for publication: Linearly convergent bilevel optimization with single-step inner methods
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6155063)
