Forward-backward truncated Newton methods for convex composite optimization

Abstract

This paper proposes two proximal Newton-CG methods for convex nonsmooth optimization problems in composite form. The algorithms are based on a a reformulation of the original nonsmooth problem as the unconstrained minimization of a continuously differentiable function, namely the forward-backward envelope (FBE). The first algorithm is based on a standard line search strategy, whereas the second one combines the global efficiency estimates of the corresponding first-order methods, while achieving fast asymptotic convergence rates. Furthermore, they are computationally attractive since each Newton iteration requires the approximate solution of a linear system of usually small dimension.

References in corpus (1)

• A quasi-Newton proximal splitting method

Cited by in corpus (12)

• A General Analysis of the Convergence of ADMM
• Embedded nonlinear model predictive control for obstacle avoidance using PANOC
• On the acceleration of forward-backward splitting via an inexact Newton method
• Proximal Gradient Algorithms: Applications in Signal Processing
• Douglas-Rachford Splitting: Complexity Estimates and Accelerated Variants
• A Moreau-Yosida approximation scheme for a class of high-dimensional posterior distributions
• A Stochastic Extra-Step Quasi-Newton Method for Nonsmooth Nonconvex Optimization
• An Inexact Semi-smooth Newton Method on Riemannian Manifolds with Application to Duality-based Total Variation Denoising
• A Trust-Region Method For Nonsmooth Nonconvex Optimization
• A trust region-type normal map-based semismooth Newton method for nonsmooth nonconvex composite optimization
• Envelope Functions: Unifications and Further Properties
• Inexact proximal DC Newton-type method for nonconvex composite functions