Theoretical Insights into Line Graph Transformation on Graph Learning

Fan Yang, Xingyue Huang

Keywords: Graph isomorphism testing, Strongly regular graphs, CFI graphs, Expressivity, Graph neural networks.

TL;DR: We show theoretical proofs relating to line graph transformations with CFI and strongly regular graphs and discuss their implication on graph learning

Abstract:

Line graph transformation has been widely studied in graph theory, where each node in a line graph corresponds to an edge in the original graph. This has inspired a series of graph neural networks (GNNs) applied on transformed line graphs, which has proven effective in various graph representation learning tasks. However, there is limited theoretical study on how line graph transformation affects expressivity of GNN models. In this study, we focus on two types of graphs known to be challenging to the Weisfeiler-Leman (WL) tests: Cai-Fürer-Immerman (CFI) graphs and strongly regular graphs, and show that applying line graph transformation helps exclude these challenging graph properties, thus potentially assist WL tests in distinguishing these graphs.
We empirically validate our findings by conducting a series of experiments that compare the accuracy and efficiency of graph isomorphism tests and GNNs on both line-transformed and original graphs across these graph structure types.

Submission Number: 22