Fast-rate PAC-Bayes generalization bounds via shifted rademacher processes

Jun Yang*, Shengyang Sun, Daniel M. Roy

*Corresponding author af dette arbejde

Publikation: Bidrag til tidsskriftKonferenceartikelForskningpeer review

18 Citationer (Scopus)

Abstract

The developments of Rademacher complexity and PAC-Bayesian theory have been largely independent. One exception is the PAC-Bayes theorem of Kakade, Sridharan, and Tewari [21], which is established via Rademacher complexity theory by viewing Gibbs classifiers as linear operators. The goal of this paper is to extend this bridge between Rademacher complexity and state-of-the-art PAC-Bayesian theory. We first demonstrate that one can match the fast rate of Catoni's PAC-Bayes bounds [8] using shifted Rademacher processes [27, 43, 44]. We then derive a new fast-rate PAC-Bayes bound in terms of the “flatness” of the empirical risk surface on which the posterior concentrates. Our analysis establishes a new framework for deriving fast-rate PAC-Bayes bounds and yields new insights on PAC-Bayesian theory.

OriginalsprogEngelsk
TidsskriftAdvances in Neural Information Processing Systems
Vol/bind32
ISSN1049-5258
StatusUdgivet - 2019
Udgivet eksterntJa
Begivenhed33rd Annual Conference on Neural Information Processing Systems, NeurIPS 2019 - Vancouver, Canada
Varighed: 8 dec. 201914 dec. 2019

Konference

Konference33rd Annual Conference on Neural Information Processing Systems, NeurIPS 2019
Land/OmrådeCanada
ByVancouver
Periode08/12/201914/12/2019
SponsorCitadel, Doc.AI, et al., Lambda, Lyft, Microsoft Research

Bibliografisk note

Publisher Copyright:
© 2019 Neural information processing systems foundation. All rights reserved.

Citationsformater