Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting
About
Modeling complex spatial and temporal correlations in the correlated time series data is indispensable for understanding the traffic dynamics and predicting the future status of an evolving traffic system. Recent works focus on designing complicated graph neural network architectures to capture shared patterns with the help of pre-defined graphs. In this paper, we argue that learning node-specific patterns is essential for traffic forecasting while the pre-defined graph is avoidable. To this end, we propose two adaptive modules for enhancing Graph Convolutional Network (GCN) with new capabilities: 1) a Node Adaptive Parameter Learning (NAPL) module to capture node-specific patterns; 2) a Data Adaptive Graph Generation (DAGG) module to infer the inter-dependencies among different traffic series automatically. We further propose an Adaptive Graph Convolutional Recurrent Network (AGCRN) to capture fine-grained spatial and temporal correlations in traffic series automatically based on the two modules and recurrent networks. Our experiments on two real-world traffic datasets show AGCRN outperforms state-of-the-art by a significant margin without pre-defined graphs about spatial connections.
Key insights
What traffic forecasting problem does AGCRN solve?
AGCRN forecasts traffic by learning node-specific patterns and spatial dependencies automatically, without requiring a pre-defined graph of road connections.
How does AGCRN differ from prior GCN traffic models?
Unlike prior GCN models sharing parameters and needing pre-defined graphs, AGCRN uses NAPL and DAGG modules. On PeMSD4/PeMSD8, it improved MAE by over 5% versus the best baseline.
What are the practical uses of these findings?
The modules are demonstrated as benchmarks on public highway datasets. The authors suggest adaptability to broader correlated time-series tasks like epidemic, meteorology, and stock forecasting.
What limitations and future work do the authors note?
AGCRN has about five times more parameters than DCRNN at embedding dimension 10. Future work includes validating scalability on more tasks and adapting modules to other GCN models.