A PAC-Bayes bound for deterministic classifiers - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Preprints, Working Papers, ... Year : 2022

A PAC-Bayes bound for deterministic classifiers

Abstract

We establish a disintegrated PAC-Bayesian bound, for classifiers that are trained via continuous-time (non-stochastic) gradient descent. Contrarily to what is standard in the PAC-Bayesian setting, our result applies to a training algorithm that is deterministic, conditioned on a random initialisation, without requiring any $\textit{de-randomisation}$ step. We provide a broad discussion of the main features of the bound that we propose, and we study analytically and empirically its behaviour on linear models, finding promising results.
Fichier principal
Vignette du fichier
2209.02525.pdf (727.81 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03815146 , version 1 (14-10-2022)

Identifiers

Cite

Eugenio Clerico, George Deligiannidis, Benjamin Guedj, Arnaud Doucet. A PAC-Bayes bound for deterministic classifiers. 2022. ⟨hal-03815146⟩
27 View
37 Download

Altmetric

Share

Gmail Facebook X LinkedIn More