REVIEW 3 cited by
Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
abstract
Backtracking linesearch is the de facto approach for minimizing continuously differentiable functions with locally Lipschitz gradient. In recent years, it has been shown that in the convex setting it is possible to avoid linesearch altogether, and to allow the stepsize to adapt based on a local smoothness estimate without any backtracks or evaluations of the function value. In this work we propose an adaptive proximal gradient method, adaPG, that uses novel estimates of the local smoothness modulus which leads to less conservative stepsize updates and that can additionally cope with nonsmooth terms. This idea is extended to the primal-dual setting where an adaptive three-term primal-dual algorithm, adaPD, is proposed which can be viewed as an extension of the PDHG method. Moreover, in this setting the "essentially" fully adaptive variant adaPD$^+$ is proposed that avoids evaluating the linear operator norm by invoking a backtracking procedure, that, remarkably, does not require extra gradient evaluations. Numerical simulations demonstrate the effectiveness of the proposed algorithms compared to the state of the art.
Forward citations
Cited by 3 Pith papers
-
Adaptive Stepsize Selection in Decentralized Convex Optimization
A fully local adaptive step-size scheme achieves linear (strongly convex) and sublinear (convex) convergence rates, matching tuned nonadaptive decentralized methods.
-
Two Adaptive Accelerated Golden Ratio Primal--Dual Algorithms With an Application to Poisson Imaging Problem
An adaptive golden-ratio primal-dual algorithm is shown to need no step-size cap or linesearch, with O(1/N) rates, plus two strongly-convex-focused variants with O(1/N²) rates.
-
Kahan's Automatic Step-Size Control for Unconstrained Optimization
Kahan's KGD step-size is shown to converge at least R-linearly with rate 1-1/cond(H) for quadratics, and an adaptive generalization for general optimization is proved and tested.
Discussion (0). Sign in to comment.