ReInc: Scaling Training of Dynamic Graph Neural Networks

Dynamic Graph Neural Networks (DGNNs) have gained widespread attention due to their applicability in diverse do- mains such as traffic network prediction, epidemiological forecasting, and social network analysis. In this paper, we present REINC, a system designed to enable efficient and scalable training of DGNNs on large-scale graphs. REINC introduces key innovations that capitalize on the unique com- bination of Graph Neural Networks (GNNs) and Recurrent Neural Networks (RNNs) inherent in DGNNs. By reusing intermediate results and incrementally computing aggrega- tions across consecutive graph snapshots, REINC significantly enhances computational efficiency. To support these optimiza- tions, REINC incorporates a novel two-level caching mecha- nism with a specialized caching policy aligned to the DGNN execution workflow. Additionally, REINC addresses the chal- lenges of managing structural and temporal dependencies in dynamic graphs through a new distributed training strategy. This approach eliminates communication overheads associ- ated with accessing remote features and redistributing inter- mediate results. Experimental results demonstrate that REINC achieves up to an order of magnitude speedup compared to state-of-the-art frameworks, tested across various dynamic GNN architectures and real-world graph datasets.