Date
2024-02-28Author
Beddar-Wiesing, SilviaD’Inverno, Giuseppe AlessioGraziani, CaterinaLachi, VeronicaMoallemy-Oureh, AliceScarselli, FrancoThomas, Josephine MariaMetadata
Show full item record
Aufsatz![](/themes/Mirage2/images/Open_Access_logo_PLoS_transparent.svg)
Weisfeiler-Lehman goes dynamic: An analysis of the expressive power of Graph Neural Networks for attributed and dynamic graphs
Abstract
Graph Neural Networks (GNNs) are a large class of relational models for graph processing. Recent theoretical studies on the expressive power of GNNs have focused on two issues. On the one hand, it has been proven that GNNs are as powerful as the Weisfeiler–Lehman test (1-WL) in their ability to distinguish graphs. Moreover, it has been shown that the equivalence enforced by 1-WL equals unfolding equivalence. On the other hand, GNNs turned out to be universal approximators on graphs modulo the constraints enforced by 1-WL/unfolding equivalence. However, these results only apply to Static Attributed Undirected Homogeneous Graphs (SAUHG) with node attributes. In contrast, real-life applications often involve a much larger variety of graph types. In this paper, we conduct a theoretical analysis of the expressive power of GNNs for two other graph domains that are particularly interesting in practical applications, namely dynamic graphs and SAUGHs with edge attributes. Dynamic graphs are widely used in modern applications; hence, the study of the expressive capability of GNNs in this domain is essential for practical reasons and, in addition, it requires a new analyzing approach due to the difference in the architecture of dynamic GNNs compared to static ones. On the other hand, the examination of SAUHGs is of particular relevance since they act as a standard form for all graph types: it has been shown that all graph types can be transformed without loss of information to SAUHGs with both attributes on nodes and edges. This paper considers generic GNN models and appropriate 1-WL tests for those domains. Then, the known results on the expressive power of GNNs are extended to the mentioned domains: it is proven that GNNs have the same capability as the 1-WL test, the 1-WL equivalence equals unfolding equivalence and that GNNs are universal approximators modulo 1-WL/unfolding equivalence. Moreover, the proof of the approximation capability is mostly constructive and allows us to deduce hints on the architecture of GNNs that can achieve the desired approximation.
Citation
In: Neural Networks Volume 173 (2024-02-28) eissn:0893-6080Sponsorship
Gefördert im Rahmen des Projekts DEALCitation
@article{doi:10.17170/kobra-202403139766,
author={Beddar-Wiesing, Silvia and D’Inverno, Giuseppe Alessio and Graziani, Caterina and Lachi, Veronica and Moallemy-Oureh, Alice and Scarselli, Franco and Thomas, Josephine Maria},
title={Weisfeiler-Lehman goes dynamic: An analysis of the expressive power of Graph Neural Networks for attributed and dynamic graphs},
journal={Neural Networks},
year={2024}
}
0500 Oax 0501 Text $btxt$2rdacontent 0502 Computermedien $bc$2rdacarrier 1100 2024$n2024 1500 1/eng 2050 ##0##http://hdl.handle.net/123456789/15551 3000 Beddar-Wiesing, Silvia 3010 D’Inverno, Giuseppe Alessio 3010 Graziani, Caterina 3010 Lachi, Veronica 3010 Moallemy-Oureh, Alice 3010 Scarselli, Franco 3010 Thomas, Josephine Maria 4000 Weisfeiler-Lehman goes dynamic: An analysis of the expressive power of Graph Neural Networks for attributed and dynamic graphs / Beddar-Wiesing, Silvia 4030 4060 Online-Ressource 4085 ##0##=u http://nbn-resolving.de/http://hdl.handle.net/123456789/15551=x R 4204 \$dAufsatz 4170 5550 {{Graph}} 5550 {{Algorithmus}} 5550 {{Neuronales Netz}} 7136 ##0##http://hdl.handle.net/123456789/15551
2024-03-13T10:52:28Z 2024-03-13T10:52:28Z 2024-02-28 doi:10.17170/kobra-202403139766 http://hdl.handle.net/123456789/15551 Gefördert im Rahmen des Projekts DEAL eng Namensnennung 4.0 International http://creativecommons.org/licenses/by/4.0/ Graph Neural Network Dynamic graphs GNN expressivity Unfolding trees Weisfeiler-Lehman test 004 600 Weisfeiler-Lehman goes dynamic: An analysis of the expressive power of Graph Neural Networks for attributed and dynamic graphs Aufsatz Graph Neural Networks (GNNs) are a large class of relational models for graph processing. Recent theoretical studies on the expressive power of GNNs have focused on two issues. On the one hand, it has been proven that GNNs are as powerful as the Weisfeiler–Lehman test (1-WL) in their ability to distinguish graphs. Moreover, it has been shown that the equivalence enforced by 1-WL equals unfolding equivalence. On the other hand, GNNs turned out to be universal approximators on graphs modulo the constraints enforced by 1-WL/unfolding equivalence. However, these results only apply to Static Attributed Undirected Homogeneous Graphs (SAUHG) with node attributes. In contrast, real-life applications often involve a much larger variety of graph types. In this paper, we conduct a theoretical analysis of the expressive power of GNNs for two other graph domains that are particularly interesting in practical applications, namely dynamic graphs and SAUGHs with edge attributes. Dynamic graphs are widely used in modern applications; hence, the study of the expressive capability of GNNs in this domain is essential for practical reasons and, in addition, it requires a new analyzing approach due to the difference in the architecture of dynamic GNNs compared to static ones. On the other hand, the examination of SAUHGs is of particular relevance since they act as a standard form for all graph types: it has been shown that all graph types can be transformed without loss of information to SAUHGs with both attributes on nodes and edges. This paper considers generic GNN models and appropriate 1-WL tests for those domains. Then, the known results on the expressive power of GNNs are extended to the mentioned domains: it is proven that GNNs have the same capability as the 1-WL test, the 1-WL equivalence equals unfolding equivalence and that GNNs are universal approximators modulo 1-WL/unfolding equivalence. Moreover, the proof of the approximation capability is mostly constructive and allows us to deduce hints on the architecture of GNNs that can achieve the desired approximation. open access Beddar-Wiesing, Silvia D’Inverno, Giuseppe Alessio Graziani, Caterina Lachi, Veronica Moallemy-Oureh, Alice Scarselli, Franco Thomas, Josephine Maria 16 Seiten doi:10.1016/j.neunet.2024.106213 Graph Algorithmus Neuronales Netz publishedVersion eissn:0893-6080 Neural Networks Volume 173 false 106213
The following license files are associated with this item: