Improving the Expressive Power of Graph Neural Network with Tinhofer Algorithm
This addresses a fundamental limitation in GNNs for graph-based data processing, though it appears incremental as it builds on existing message-passing schemes.
The paper tackles the limited expressive power of Graph Neural Networks (GNNs) by proposing WLT-GNN, a variation based on the Tinhofer algorithm, which theoretically breaks through the Weisfeiler-Lehman test limitation and shows comparable performance and better expressive power on several datasets.
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.