<div class="csl-bib-body">
<div class="csl-entry">Welke, P., Thiessen, M., & Gärtner, T. (2022, November 30). <i>Expectation Complete Graph Representations Using Graph Homomorphisms</i> [Poster Presentation]. First Learning on Graphs Conference (LoG 2022), Unknown. https://doi.org/10.34726/3883</div>
</div>
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/175918
-
dc.identifier.uri
https://doi.org/10.34726/3883
-
dc.description.abstract
We propose and study a practical graph embedding that *in expectation* is able to distinguish all non-isomorphic graphs and can be computed in polynomial time. The embedding is based on Lovász' characterisation of graph isomorphism through an infinite dimensional vector of homomorphism counts. Recent work has studied the expressiveness of graph embeddings by comparing their ability to distinguish graphs to that of the Weisfeiler-Leman hierarchy. While previous methods have either limited expressiveness or are computationally impractical, we devise efficient sampling-based alternatives that are maximally expressive in expectation. We empirically evaluate our proposed embeddings and show competitive results on several benchmark graph learning tasks.
en
dc.language.iso
en
-
dc.rights.uri
http://creativecommons.org/licenses/by/4.0/
-
dc.subject
Machine Learning
en
dc.title
Expectation Complete Graph Representations Using Graph Homomorphisms