<div class="csl-bib-body">
<div class="csl-entry">Jogl, F., Welke, P., & Gärtner, T. (2024). Is Expressivity Essential for the Predictive Performance of Graph Neural Networks? In <i>NeurIPS 2024 Workshop on Scientific Methods for Understanding Deep Learning</i>. NeurIPS 2024 The Thirty-Eighth Annual Conference on Neural Information Processing Systems, Vancouver, Canada.</div>
</div>
Motivated by the large amount of research on the expressivity of GNNs, we study the impact of expressivity on the predictive performance of GNNs. By performing knowledge distillation from highly expressive teacher GNNs to less expressive
student GNNs, we demonstrate that knowledge distillation reduces the predictive performance gap between teachers and students significantly. As knowledge distillation does not increase the expressivity of the student GNN, it follows that most of this gap in predictive performance cannot be due to expressivity.
en
dc.description.sponsorship
WWTF Wiener Wissenschafts-, Forschu und Technologiefonds
-
dc.language.iso
en
-
dc.subject
Graph Neural Networks
en
dc.subject
Expressivity
en
dc.subject
Machine Learning
en
dc.title
Is Expressivity Essential for the Predictive Performance of Graph Neural Networks?
en
dc.type
Inproceedings
en
dc.type
Konferenzbeitrag
de
dc.relation.publication
NeurIPS 2024 Workshop on Scientific Methods for Understanding Deep Learning
-
dc.relation.grantno
ICT22-059
-
dc.type.category
Full-Paper Contribution
-
tuw.booktitle
NeurIPS 2024 Workshop on Scientific Methods for Understanding Deep Learning
-
tuw.peerreviewed
true
-
tuw.project.title
Structured Data Learning with Generalized Similarities
-
tuw.researchTopic.id
I4
-
tuw.researchTopic.name
Information Systems Engineering
-
tuw.researchTopic.value
100
-
tuw.publication.orgunit
E192-02 - Forschungsbereich Databases and Artificial Intelligence
-
tuw.publication.orgunit
E194-06 - Forschungsbereich Machine Learning
-
tuw.publication.orgunit
E056-10 - Fachbereich SecInt-Secure and Intelligent Human-Centric Digital Technologies
-
tuw.publication.orgunit
E056-23 - Fachbereich Innovative Combinations and Applications of AI and ML (iCAIML)
-
tuw.publication.orgunit
E056-26 - Fachbereich Automated Reasoning
-
dc.description.numberOfPages
10
-
tuw.author.orcid
0000-0002-2123-3781
-
tuw.author.orcid
0000-0001-5985-9213
-
tuw.event.name
NeurIPS 2024 The Thirty-Eighth Annual Conference on Neural Information Processing Systems
en
tuw.event.startdate
10-12-2024
-
tuw.event.enddate
15-12-2024
-
tuw.event.online
On Site
-
tuw.event.type
Event for scientific audience
-
tuw.event.place
Vancouver
-
tuw.event.country
CA
-
tuw.event.presenter
Welke, Pascal
-
tuw.event.track
Single Track
-
wb.sciencebranch
Informatik
-
wb.sciencebranch.oefos
1020
-
wb.sciencebranch.value
100
-
item.openairetype
conference paper
-
item.cerifentitytype
Publications
-
item.grantfulltext
none
-
item.languageiso639-1
en
-
item.openairecristype
http://purl.org/coar/resource_type/c_5794
-
item.fulltext
no Fulltext
-
crisitem.project.funder
WWTF Wiener Wissenschafts-, Forschu und Technologiefonds
-
crisitem.project.grantno
ICT22-059
-
crisitem.author.dept
E192-02 - Forschungsbereich Databases and Artificial Intelligence
-
crisitem.author.dept
E194-06 - Forschungsbereich Machine Learning
-
crisitem.author.dept
E194-06 - Forschungsbereich Machine Learning
-
crisitem.author.orcid
0000-0002-2123-3781
-
crisitem.author.orcid
0000-0001-5985-9213
-
crisitem.author.parentorg
E192 - Institut für Logic and Computation
-
crisitem.author.parentorg
E194 - Institut für Information Systems Engineering
-
crisitem.author.parentorg
E194 - Institut für Information Systems Engineering
