We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:


References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Improving the Expressive Power of Graph Neural Network with Tinhofer Algorithm

Abstract: In recent years, Graph Neural Network (GNN) has bloomly progressed for its power in processing graph-based data. Most GNNs follow a message passing scheme, and their expressive power is mathematically limited by the discriminative ability of the Weisfeiler-Lehman (WL) test. Following Tinhofer's research on compact graphs, we propose a variation of the message passing scheme, called the Weisfeiler-Lehman-Tinhofer GNN (WLT-GNN), that theoretically breaks through the limitation of the WL test. In addition, we conduct comparative experiments and ablation studies on several well-known datasets. The results show that the proposed methods have comparable performances and better expressive power on these datasets.
Subjects: Machine Learning (cs.LG)
Cite as: arXiv:2104.01848 [cs.LG]
  (or arXiv:2104.01848v1 [cs.LG] for this version)

Submission history

From: Alan JiaXiang Guo [view email]
[v1] Mon, 5 Apr 2021 10:54:22 GMT (1188kb,D)

Link back to: arXiv, form interface, contact.