Convergence of Manifold Filter-Combine Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Johnson, David R., Chew, Joyce, Viswanath, Siddharth, De Brouwer, Edward, Needell, Deanna, Krishnaswamy, Smita, Perlmutter, Michael
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910656927629312
author Johnson, David R.
Chew, Joyce
Viswanath, Siddharth
De Brouwer, Edward
Needell, Deanna
Krishnaswamy, Smita
Perlmutter, Michael
author_facet Johnson, David R.
Chew, Joyce
Viswanath, Siddharth
De Brouwer, Edward
Needell, Deanna
Krishnaswamy, Smita
Perlmutter, Michael
contents In order to better understand manifold neural networks (MNNs), we introduce Manifold Filter-Combine Networks (MFCNs). The filter-combine framework parallels the popular aggregate-combine paradigm for graph neural networks (GNNs) and naturally suggests many interesting families of MNNs which can be interpreted as the manifold analog of various popular GNNs. We then propose a method for implementing MFCNs on high-dimensional point clouds that relies on approximating the manifold by a sparse graph. We prove that our method is consistent in the sense that it converges to a continuum limit as the number of data points tends to infinity.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14639
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Convergence of Manifold Filter-Combine Networks
Johnson, David R.
Chew, Joyce
Viswanath, Siddharth
De Brouwer, Edward
Needell, Deanna
Krishnaswamy, Smita
Perlmutter, Michael
Machine Learning
Signal Processing
In order to better understand manifold neural networks (MNNs), we introduce Manifold Filter-Combine Networks (MFCNs). The filter-combine framework parallels the popular aggregate-combine paradigm for graph neural networks (GNNs) and naturally suggests many interesting families of MNNs which can be interpreted as the manifold analog of various popular GNNs. We then propose a method for implementing MFCNs on high-dimensional point clouds that relies on approximating the manifold by a sparse graph. We prove that our method is consistent in the sense that it converges to a continuum limit as the number of data points tends to infinity.
title Convergence of Manifold Filter-Combine Networks
topic Machine Learning
Signal Processing
url https://arxiv.org/abs/2410.14639