Random Propagations in GNNs

Thu Bui, Anugunj Naman, Carola-Bibiane Schönlieb, Bruno Ribeiro, Beatrice Bevilacqua, Moshe Eliasof

Track: Extended Abstract Track

Keywords: GNN, Randomness

TL;DR: We present RAP-GNN, a framework that uses randomly sampled weights and pretrained embeddings to cut training time by 58% while achieving competitive accuracy in node and graph classification tasks across diverse datasets.

Abstract: Graph learning benefits many fields. However, Graph Neural Networks (GNNs) often struggle with scalability, especially on large graphs. At the same time, many tasks seem to be simple in terms of learning, e.g., simple diffusion yields favorable performance. In this paper, we present Random Propagation GNN (RAP-GNN), a framework that addresses two main research questions: (i) can random propagations in GNNs be as effective as end-to-end optimized GNNs? and (ii) can they reduce the computational burden required by traditional GNNs? Our empirical findings indicate that RAP-GNN reduces training time by up to 58\%, while maintaining strong accuracy for node and graph classification tasks.

Submission Number: 53