Region-free Safe Screening Tests for l1-penalized Convex Problems - Archive ouverte HAL Access content directly
Conference Papers Year : 2022

Region-free Safe Screening Tests for l1-penalized Convex Problems

Abstract

We address the problem of safe screening for l1-penalized convex regression/classification problems, i.e., the identification of zero coordinates of the solutions. Unlike previous contributions of the literature, we propose a screening methodology which does not require the knowledge of a so-called "safe region". Our approach does not rely on any other assumption than convexity (in particular, no strong-convexity hypothesis is needed) and therefore applies to a wide family of convex problems. When the Fenchel conjugate of the data-fidelity term is strongly convex, we show that the popular "GAP sphere test" proposed by Fercoq et al. can be recovered as a particular case of our methodology (up to a minor modification). We illustrate numerically the performance of our procedure on the "sparse support vector machine classification" problem.
Fichier principal
Vignette du fichier
main.pdf (343.62 Ko) Télécharger le fichier

Dates and versions

hal-03806099 , version 1 (07-10-2022)

Identifiers

  • HAL Id : hal-03806099 , version 1

Cite

Cédric Herzet, Clément Elvira, Hong-Phuong Dang. Region-free Safe Screening Tests for l1-penalized Convex Problems. Eusipco 2022 - 30th European Signal Processing Conference, Aug 2022, Belgrade, Serbia. pp.1-5. ⟨hal-03806099⟩
104 View
78 Download

Share

Gmail Facebook X LinkedIn More