On Margins and Derandomisation in PAC-Bayes - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2022

On Margins and Derandomisation in PAC-Bayes

Abstract

We give a general recipe for derandomising PAC-Bayesian bounds using margins, with the critical ingredient being that our randomised predictions concentrate around some value. The tools we develop traightforwardly lead to margin bounds for various classifiers, including linear prediction—a class that includes boosting and the support vector machine—single-hidden-layer neural networks with an unusual erf activation function, and deep ReLU networks. Further, we extend to partially-derandomised predictors where only some of the randomness is removed, letting us extend bounds to cases where the concentration properties of our predictors are otherwise poor.
Fichier principal
Vignette du fichier
2107.03955.pdf (943.97 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03282597 , version 1 (09-07-2021)
hal-03282597 , version 2 (24-02-2022)

Identifiers

Cite

Felix Biggs, Benjamin Guedj. On Margins and Derandomisation in PAC-Bayes. AISTATS 2022 - 25th International Conference on Artificial Intelligence and Statistics, Mar 2022, Valencia, Spain. ⟨hal-03282597v2⟩
56 View
187 Download

Altmetric

Share

Gmail Facebook X LinkedIn More