PAC-Bayesian Margin Bounds for Convolutional Neural Networks

Recently the generalisation error of deep neural networks has been analysed through the PAC-Bayesian framework, for the case of fully connected layers. We adapt this approach to the convolutional setting.

cs.LG stat.ML Computer Science Mathematics

View graph Source

Paper

References (5)

01A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural NetworksBehnam Neyshabur, Srinadh Bhojanapalli, David A. McAllester et al.2017 · International Conference on Learning Representations · 679 citations In Library

02Nearly-tight VC-dimension bounds for piecewise linear neural networksNick Harvey, Christopher Liaw, Abbas Mehrabian2017 · Annual Conference Computational Learning Theory · 190 citations

03Sharp nonasymptotic bounds on the norm of random matrices with independent entriesA. Bandeira, R. Handel2014 · 303 citations

04Introduction to the non-asymptotic analysis of random matricesR. Vershynin2010 · Compressed Sensing · 3.1k citations

05User-Friendly Tail Bounds for Sums of Random MatricesJ. Tropp2010 · Foundations of Computational Mathematics · 1.9k citations

Similar papers

A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural NetworksBehnam Neyshabur, Srinadh Bhojanapalli, David A. McAllester et al.2017 · International Conference on Learning Representations · 677 citations In Library

On the Importance of Gradient Norm in PAC-Bayesian BoundsItai Gat, Yossi Adi, Alexander G. Schwing et al.2022 · arXiv (Cornell University) · 7 citations In Library

PAC-Bayesian risk bounds for fully connected deep neural network with Gaussian priorsThe Tien Mai2025 · 5 citations In Library

Generalization bounds for deep convolutional neural networksPhilip M. Long, Hanie Sedghi2020 · International Conference on Learning Representations In Library

Generalization bounds for deep convolutional neural networksPhilip M. Long, Hanie Sedghi2019 · International Conference on Learning Representations · 102 citations In Library

Norm-based generalisation bounds for multi-class convolutional neural networksAntoine Ledent, Yunwen Lei, M. Kloft2019 · arXiv.org · 5 citations In Library

Norm-Based Generalisation Bounds for Deep Multi-Class Convolutional Neural NetworksAntoine Ledent, Waleed Mustafa, Yunwen Lei et al.2021 · AAAI Conference on Artificial Intelligence · 24 citations In Library

Robust Large Margin Deep Neural NetworksJure Sokolić, R. Giryes, G. Sapiro et al.2016 · IEEE Transactions on Signal Processing · 349 citations In Library

PAC-Bayes Compression Bounds So Tight That They Can Explain GeneralizationSanae Lotfi, Marc Finzi, Sanyam Kapoor et al.2022 · arXiv (Cornell University) · 91 citations In Library

A PAC-Bayesian Approach to Generalization Bounds for Graph Neural NetworksRenjie Liao, Raquel Urtasun, Richard S. Zemel2020 · arXiv (Cornell University) · 113 citations In Library

Source of record

arXiv

44B

Welcome.

This panel sticks with you. Pick anything below and it opens right beside it, so you can dig through 60,000-plus records without ever losing your spot here.

Travel 44B

LibraryPapers, policy, standards, statute LabsEvery organization building AI ModelsIntelligence, price, and speed BenchmarksThe evaluation catalog IncidentsAI failures and harms SearchOne field across everything DashboardThe state of AI accountability in NY 44B RegistryThe Art. 44-B compliance portal