GIDN: A Lightweight Graph Inception Diffusion Network for High-efficient Link Prediction
This work addresses the problem of efficient link prediction for graph-based applications, presenting an incremental improvement over existing methods.
The paper tackles link prediction in graphs by proposing GIDN, a lightweight model that generalizes graph diffusion across feature spaces and uses inception modules to reduce computational costs, achieving an 11% higher performance than AGDN on the ogbl-collab dataset.
In this paper, we propose a Graph Inception Diffusion Networks(GIDN) model. This model generalizes graph diffusion in different feature spaces, and uses the inception module to avoid the large amount of computations caused by complex network structures. We evaluate GIDN model on Open Graph Benchmark(OGB) datasets, reached an 11% higher performance than AGDN on ogbl-collab dataset.