Boosted Stochastic Frank-Wolfe for Constrained Nonconvex Optimization

Navil Nandhan, Abbas Khademi, Antonio Silveti-Falls · 2026

Read Boosted Stochastic Frank-Wolfe for Constrained Nonconvex Optimization by Navil Nandhan, Abbas Khademi, Antonio Silveti-Falls (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-25255.txt.

Description

Navil Nandhan, Abbas Khademi, and Antonio Silveti-Falls extend boosted Frank-Wolfe to stochastic, constrained, nonconvex problems. Boosting aligns updates with the gradient, but prior analysis covered only convex cases. The team devises a step-size strategy independent of the gradient's Lipschitz constant, compatible with estimators from SAGA to heavy-ball momentum and zeroth-order methods. Their analysis keeps stochastic Frank-Wolfe rates and gives first boosted rates for nonconvex and quasar-convex objectives. Experiments on logistic regression and quantum tomography show faster convergence per call and in wall-clock.

Catalog record

CREATORNavil Nandhan, Abbas Khademi, Antonio Silveti-Falls
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE168350
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.25255

Browse the full library · About the Tower of Babel