Subgradient sampling for nonsmooth nonconvex minimization - IRIT - Université Toulouse 1 Capitole Access content directly
Preprints, Working Papers, ... Year : 2022

Subgradient sampling for nonsmooth nonconvex minimization


Risk minimization for nonsmooth nonconvex problems naturally leads to firstorder sampling or, by an abuse of terminology, to stochastic subgradient descent. We establish the convergence of this method in the path-differentiable case, and describe more precise results under additional geometric assumptions. We recover and improve results from Ermoliev-Norkin by using a different approach: conservative calculus and the ODE method. In the definable case, we show that first-order subgradient sampling avoids artificial critical point with probability one and applies moreover to a large range of risk minimization problems in deep learning, based on the backpropagation oracle. As byproducts of our approach, we obtain several results on integration of independent interest, such as an interchange result for conservative derivatives and integrals, or the definability of set-valued parameterized integrals.
Fichier principal
Vignette du fichier
subgradientSampling.pdf (431.46 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03579383 , version 1 (18-02-2022)
hal-03579383 , version 2 (25-02-2022)
hal-03579383 , version 3 (28-10-2022)
hal-03579383 , version 4 (26-01-2023)
hal-03579383 , version 5 (09-03-2023)


  • HAL Id : hal-03579383 , version 1


Jérôme Bolte, Tam Le, Edouard Pauwels. Subgradient sampling for nonsmooth nonconvex minimization. 2022. ⟨hal-03579383v1⟩
511 View
301 Download


Gmail Mastodon Facebook X LinkedIn More