On the VC dimension of deep group convolutional neural networks

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Sepliarskaia, Anna, Langer, Sophie, Schmidt-Hieber, Johannes
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913557399994368
author Sepliarskaia, Anna
Langer, Sophie
Schmidt-Hieber, Johannes
author_facet Sepliarskaia, Anna
Langer, Sophie
Schmidt-Hieber, Johannes
contents We study the generalization capabilities of Group Convolutional Neural Networks (GCNNs) with ReLU activation function by deriving upper and lower bounds for their Vapnik-Chervonenkis (VC) dimension. Specifically, we analyze how factors such as the number of layers, weights, and input dimension affect the VC dimension. We further compare the derived bounds to those known for other types of neural networks. Our findings extend previous results on the VC dimension of continuous GCNNs with two layers, thereby providing new insights into the generalization properties of GCNNs, particularly regarding the dependence on the input resolution of the data.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15800
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle On the VC dimension of deep group convolutional neural networks
Sepliarskaia, Anna
Langer, Sophie
Schmidt-Hieber, Johannes
Machine Learning
Statistics Theory
We study the generalization capabilities of Group Convolutional Neural Networks (GCNNs) with ReLU activation function by deriving upper and lower bounds for their Vapnik-Chervonenkis (VC) dimension. Specifically, we analyze how factors such as the number of layers, weights, and input dimension affect the VC dimension. We further compare the derived bounds to those known for other types of neural networks. Our findings extend previous results on the VC dimension of continuous GCNNs with two layers, thereby providing new insights into the generalization properties of GCNNs, particularly regarding the dependence on the input resolution of the data.
title On the VC dimension of deep group convolutional neural networks
topic Machine Learning
Statistics Theory
url https://arxiv.org/abs/2410.15800