A Novel Integer Linear Programming Approach for Global L0 Minimization - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles Journal of Machine Learning Research Year : 2023

A Novel Integer Linear Programming Approach for Global L0 Minimization

Diego Delle Donne
  • Function : Author
  • PersonId : 1118880
Matthieu Kowalski

Abstract

Given a vector y ∈ R n and a matrix H ∈ R n×m , the sparse approximation problem P 0/p asks for a point x such that ∥y-Hx∥ p ≤ α, for a given scalar α, minimizing the size of the support ∥x∥ 0 := #{j | x j ̸ = 0}. Existing convex mixed-integer programming formulations for P 0/p are of a kind referred to as "big-M ", meaning that they involve the use of a bound M on the values of x. When a proper value for M is not known beforehand, these formulations are not exact, in the sense that they may fail to recover the wanted global minimizer. In this work, we study the polytopes arising from these formulations and derive valid inequalities for them. We first use these inequalities to design a branch-and-cut algorithm for these models. Additionally, we prove that these inequalities are sufficient to describe the set of feasible supports for P 0/p. Based on this result, we introduce a new (and the first to our knowledge) M-independent integer linear programming formulation for P 0/p , which guarantees the recovery of the global minimizer. We propose a practical approach to tackle this formulation, which has exponentially many constraints. The proposed methods are then compared in computational experimentation to test their potential practical contribution.
Fichier principal
Vignette du fichier
setcov_paper_final.pdf (465.85 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04372309 , version 1 (04-01-2024)

Identifiers

  • HAL Id : hal-04372309 , version 1

Cite

Diego Delle Donne, Matthieu Kowalski, Leo Liberti. A Novel Integer Linear Programming Approach for Global L0 Minimization. Journal of Machine Learning Research, In press. ⟨hal-04372309⟩
167 View
181 Download

Share

Gmail Mastodon Facebook X LinkedIn More