Dynamic Spatiotemporal Graph Neural Network with Tensor Network
This work addresses dynamic graph modeling for time series problems, such as traffic prediction, with incremental improvements in accuracy and efficiency.
The authors tackled the challenge of dynamic spatial graph construction in GNNs for time series data by generating spatial and temporal tensor graphs to capture dynamic relations and latent patterns, optimizing them with PEPS, and achieved improved accuracy and time efficiency compared to state-of-the-art methods on public traffic datasets.
Dynamic spatial graph construction is a challenge in graph neural network (GNN) for time series data problems. Although some adaptive graphs are conceivable, only a 2D graph is embedded in the network to reflect the current spatial relation, regardless of all the previous situations. In this work, we generate a spatial tensor graph (STG) to collect all the dynamic spatial relations, as well as a temporal tensor graph (TTG) to find the latent pattern along time at each node. These two tensor graphs share the same nodes and edges, which leading us to explore their entangled correlations by Projected Entangled Pair States (PEPS) to optimize the two graphs. We experimentally compare the accuracy and time costing with the state-of-the-art GNN based methods on the public traffic datasets.