Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jiang, Qin, Wang, Chengjia, Lones, Michael, Pang, Wei
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2502.00140
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909471841714176
author Jiang, Qin
Wang, Chengjia
Lones, Michael
Pang, Wei
author_facet Jiang, Qin
Wang, Chengjia
Lones, Michael
Pang, Wei
contents While Graph Neural Networks (GNNs) have achieved remarkable success, their design largely relies on empirical intuition rather than theoretical understanding. In this paper, we present a comprehensive analysis of GNN behavior through three fundamental aspects: (1) we establish that \textbf{$k$-layer} Message Passing Neural Networks efficiently aggregate \textbf{$k$-hop} neighborhood information through iterative computation, (2) analyze how different loop structures influence neighborhood computation, and (3) examine behavior across structure-feature hybrid and structure-only tasks. For deeper GNNs, we demonstrate that gradient-related issues, rather than just over-smoothing, can significantly impact performance in sparse graphs. We also analyze how different normalization schemes affect model performance and how GNNs make predictions with uniform node features, providing a theoretical framework that bridges the gap between empirical success and theoretical understanding.
format Preprint
id arxiv_https___arxiv_org_abs_2502_00140
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Demystifying MPNNs: Message Passing as Merely Efficient Matrix Multiplication
Jiang, Qin
Wang, Chengjia
Lones, Michael
Pang, Wei
Machine Learning
Artificial Intelligence
Neural and Evolutionary Computing
Social and Information Networks
While Graph Neural Networks (GNNs) have achieved remarkable success, their design largely relies on empirical intuition rather than theoretical understanding. In this paper, we present a comprehensive analysis of GNN behavior through three fundamental aspects: (1) we establish that \textbf{$k$-layer} Message Passing Neural Networks efficiently aggregate \textbf{$k$-hop} neighborhood information through iterative computation, (2) analyze how different loop structures influence neighborhood computation, and (3) examine behavior across structure-feature hybrid and structure-only tasks. For deeper GNNs, we demonstrate that gradient-related issues, rather than just over-smoothing, can significantly impact performance in sparse graphs. We also analyze how different normalization schemes affect model performance and how GNNs make predictions with uniform node features, providing a theoretical framework that bridges the gap between empirical success and theoretical understanding.
title Demystifying MPNNs: Message Passing as Merely Efficient Matrix Multiplication
topic Machine Learning
Artificial Intelligence
Neural and Evolutionary Computing
Social and Information Networks
url https://arxiv.org/abs/2502.00140