The reviewed record of science sign in
Pith

arxiv: 2102.08649 · v3 · pith:VQW4EKFH · submitted 2021-02-17 · stat.ML · cs.LG

A General Framework for the Practical Disintegration of PAC-Bayesian Bounds

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:VQW4EKFHrecord.jsonopen to challenge →

classification stat.ML cs.LG
keywords boundspac-bayesianframeworkgeneralizationnetworksneuralpracticalstep
0
0 comments X
read the original abstract

PAC-Bayesian bounds are known to be tight and informative when studying the generalization ability of randomized classifiers. However, they require a loose and costly derandomization step when applied to some families of deterministic models such as neural networks. As an alternative to this step, we introduce new PAC-Bayesian generalization bounds that have the originality to provide disintegrated bounds, i.e., they give guarantees over one single hypothesis instead of the usual averaged analysis. Our bounds are easily optimizable and can be used to design learning algorithms. We illustrate this behavior on neural networks, and we show a significant practical improvement over the state-of-the-art framework.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels

    cs.LG 2026-05 unverdicted novelty 7.0

    Symmetrization of multi-class losses produces a unique convex symmetric loss that locally approximates others and supports robust neural training under label noise.

  2. Smoothness-Based Derandomization of PAC-Bayes Bounds

    cs.LG 2026-06 unverdicted novelty 6.0

    Derives smoothness-based PAC-Bayes bounds for deterministic predictors by bounding the Jensen gap class via Rademacher complexity, yielding flatness terms in Jacobians/Hessians, and proposes a corresponding regularize...

  3. Smoothness-Based Derandomization of PAC-Bayes Bounds

    cs.LG 2026-06 unverdicted novelty 6.0

    Derives smoothness-based PAC-Bayes derandomization bounds for deterministic predictors using Rademacher complexity of the Jensen gap class, yielding Jacobian/Hessian flatness terms and a practical regularizer tested o...

  4. Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels

    cs.LG 2026-05 unverdicted novelty 6.0

    Symmetrizing cross-entropy produces the unique convex multi-class unhinged loss, which locally approximates other symmetric losses, and enables new interpolating losses SGCE and alpha-MAE with competitive performance ...