Journal ArticleDOI
Some comments of Wolfe's `away step'
J Guélat,Patrice Marcotte +1 more
TLDR
It is given a detailed proof, under slightly weaker conditions on the objective function, that a modified Frank-Wolfe algorithm based on Wolfe's ‘away step’ strategy can achieve geometric convergence, provided a strict complementarity assumption holds.Abstract:
We give a detailed proof, under slightly weaker conditions on the objective function, that a modified Frank-Wolfe algorithm based on Wolfe's ‘away step’ strategy can achieve geometric convergence, provided a strict complementarity assumption holds.read more
Citations
More filters
Posted Content
Boosting Variational Inference: an Optimization Perspective
TL;DR: In this article, the convergence properties of boosting variational inference were studied from a modern optimization viewpoint by establishing connections to the classic Frank-Wolfe algorithm, and the sufficient conditions for convergence, explicit rates, and algorithmic simplifications.
Proceedings Article
Boosting Frank-Wolfe by Chasing Gradients
TL;DR: In this article, the authors propose to align the descent direction with that of the negative gradient via a subroutine to speed up the Frank-Wolfe algorithm and derive convergence rates of O(1/t) to O(e − ϵ t ) for the first-order optimization algorithm.
Journal ArticleDOI
Models and Software for Urban and Regional Transportation Planning: The Contributions of the Center for Research on Transportation
TL;DR: A semi-technical and somewhat journalistic account of the contributions to the methods used for quantitative transportation planning by professors, researchers and graduate students who have been active at the Centre for Research on Transportation of the University of Montreal since its inception.
Journal ArticleDOI
Multi-label core vector machine with a zero label
TL;DR: This paper extends Rank-CVM via adding a zero label to construct its variant with azero label, which is formulated as the same quadratic programming form with a unit simplex constraint and non-negative ones as Rank- CVM, and then is solved by Frank–Wolfe method efficiently.
Journal ArticleDOI
First-order Methods for the Impatient: Support Identification in Finite Time with Convergent Frank--Wolfe Variants
TL;DR: This paper analyzes two well-known and widely used variants of the Frank--Wolfe algorithm and first proves the existence of a nonconvex function over the unit simplex.
References
More filters