Linearly-convergent FISTA variant for composite optimization with duality
From MaRDI portal
Publication:6101606
Abstract: Many large-scale optimization problems can be expressed as composite optimization models. Accelerated first-order methods such as the fast iterative shrinkage-thresholding algorithm (FISTA) have proven effective for numerous large composite models. In this paper, we present a new variation of FISTA, to be called C-FISTA, which obtains global linear convergence for a broader class of composite models than many of the latest FISTA variants. We demonstrate the versatility and effectiveness of C-FISTA by showing C-FISTA outperforms current first-order solvers on both group Lasso and group logistic regression models. Furthermore, we utilize Fenchel duality to prove C-FISTA provides global linear convergence for a large class of convex models without the loss of global linear convergence.
Recommendations
- Another look at the fast iterative shrinkage/thresholding algorithm (FISTA)
- A fast dual proximal gradient algorithm for convex minimization and applications
- Stochastic primal dual fixed point method for composite optimization
- Improving ``fast iterative shrinkage-thresholding algorithm: faster, smarter, and greedier
- A nested primal-dual FISTA-like scheme for composite convex optimization problems
Cites work
- A cyclic projection algorithm via duality
- A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
- A Gauss-Newton method for convex composite optimization
- A Generalized Accelerated Composite Gradient Method: Uniting Nesterov's Fast Gradient Method and FISTA
- A Parallel Algorithm for a Class of Convex Programs
- A parallel descent algorithm for convex programming
- A tutorial on geometric programming
- A unified approach to error bounds for structured convex optimization problems
- An Accelerated Composite Gradient Method for Large-Scale Composite Objective Problems
- An efficient Hessian based algorithm for solving large-scale sparse group Lasso problems
- An introduction to continuous optimization for imaging
- Asymptotic properties of the Fenchel dual functional and applications to decomposition problems
- Backtracking strategies for accelerated descent methods with smooth composite objectives
- Convergence of an Inexact Algorithm for Composite Nonsmooth Optimization
- Convex Analysis
- Efficiency of minimizing compositions of convex functions and smooth maps
- Efficient block-coordinate descent algorithms for the group Lasso
- Error bounds, quadratic growth, and linear convergence of proximal methods
- Gradient methods for minimizing composite functions
- scientific article; zbMATH DE number 845714 (Why is no real title available?)
- Katyusha: the first direct acceleration of stochastic gradient methods
- Linear convergence of first order methods for non-strongly convex optimization
- Linearly convergent away-step conditional gradient for non-strongly convex functions
- Model Selection and Estimation in Regression with Grouped Variables
- On the linear convergence of the alternating direction method of multipliers
- Scaled, inexact, and adaptive generalized FISTA for strongly convex optimization
- The Group Lasso for Logistic Regression
Cited in
(4)- A C-FISTA-type Krasnosel'skiĭ-Mann iteration with applications to TV-based denoising problem and image in-painting
- C-FISTA type projection algorithm for quasi-variational inequalities
- Forward-backward-forward algorithm with golden ratio momentum
- A C-FISTA-type proximal point algorithm for strongly quasiconvex pseudomonotone equilibrium problems
This page was built for publication: Linearly-convergent FISTA variant for composite optimization with duality
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6101606)
