Classifying Nodes in Graphs without GNNs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Winter, Daniel, Cohen, Niv, Hoshen, Yedid
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917585189076992
author Winter, Daniel
Cohen, Niv
Hoshen, Yedid
author_facet Winter, Daniel
Cohen, Niv
Hoshen, Yedid
contents Graph neural networks (GNNs) are the dominant paradigm for classifying nodes in a graph, but they have several undesirable attributes stemming from their message passing architecture. Recently, distillation methods succeeded in eliminating the use of GNNs at test time but they still require them during training. We perform a careful analysis of the role that GNNs play in distillation methods. This analysis leads us to propose a fully GNN-free approach for node classification, not requiring them at train or test time. Our method consists of three key components: smoothness constraints, pseudo-labeling iterations and neighborhood-label histograms. Our final approach can match the state-of-the-art accuracy on standard popular benchmarks such as citation and co-purchase networks, without training a GNN.
format Preprint
id arxiv_https___arxiv_org_abs_2402_05934
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Classifying Nodes in Graphs without GNNs
Winter, Daniel
Cohen, Niv
Hoshen, Yedid
Machine Learning
Social and Information Networks
Graph neural networks (GNNs) are the dominant paradigm for classifying nodes in a graph, but they have several undesirable attributes stemming from their message passing architecture. Recently, distillation methods succeeded in eliminating the use of GNNs at test time but they still require them during training. We perform a careful analysis of the role that GNNs play in distillation methods. This analysis leads us to propose a fully GNN-free approach for node classification, not requiring them at train or test time. Our method consists of three key components: smoothness constraints, pseudo-labeling iterations and neighborhood-label histograms. Our final approach can match the state-of-the-art accuracy on standard popular benchmarks such as citation and co-purchase networks, without training a GNN.
title Classifying Nodes in Graphs without GNNs
topic Machine Learning
Social and Information Networks
url https://arxiv.org/abs/2402.05934