دورية أكاديمية

A Frobenius Norm Regularization Method for Convolutional Kernel Tensors in Neural Networks.

التفاصيل البيبلوغرافية
العنوان: A Frobenius Norm Regularization Method for Convolutional Kernel Tensors in Neural Networks.
المؤلفون: Guo PC; School of Science, China University of Geosciences, Beijing 100083, China.
المصدر: Computational intelligence and neuroscience [Comput Intell Neurosci] 2022 Aug 25; Vol. 2022, pp. 3277730. Date of Electronic Publication: 2022 Aug 25 (Print Publication: 2022).
نوع المنشور: Journal Article
اللغة: English
بيانات الدورية: Publisher: Hindawi Pub. Corp Country of Publication: United States NLM ID: 101279357 Publication Model: eCollection Cited Medium: Internet ISSN: 1687-5273 (Electronic) NLM ISO Abbreviation: Comput Intell Neurosci Subsets: MEDLINE
أسماء مطبوعة: Original Publication: New York, NY : Hindawi Pub. Corp.
مواضيع طبية MeSH: Neural Networks, Computer* , Algorithms*
مستخلص: The convolutional neural network is a very important model of deep learning. It can help avoid the exploding/vanishing gradient problem and improve the generalizability of a neural network if the singular values of the Jacobian of a layer are bounded around 1 in the training process. We propose a new Frobenius norm penalty function for a convolutional kernel tensor to let the singular values of the corresponding transformation matrix be bounded around 1. We show how to carry out the gradient-type methods. This provides a potentially useful regularization method for the weights of convolutional layers.
Competing Interests: There are no conflicts of interest to declare.
(Copyright © 2022 Pei-Chang Guo.)
References: Nature. 2015 May 28;521(7553):436-44. (PMID: 26017442)
IEEE Trans Magn. 2015 Feb;51(2):1-4. (PMID: 26203196)
IEEE Trans Cybern. 2020 Jul;50(7):3033-3044. (PMID: 31021810)
IEEE Trans Cybern. 2018 Feb;48(2):648-660. (PMID: 28237940)
IEEE Trans Neural Netw Learn Syst. 2020 May;31(5):1747-1756. (PMID: 31329134)
تواريخ الأحداث: Date Created: 20221229 Date Completed: 20221230 Latest Revision: 20230103
رمز التحديث: 20240628
مُعرف محوري في PubMed: PMC9792232
DOI: 10.1155/2022/3277730
PMID: 36579174
قاعدة البيانات: MEDLINE
الوصف
تدمد:1687-5273
DOI:10.1155/2022/3277730