About this document

DSTF-Decoupled Dynamic Spatial-Temporal Graph Neural Network For Traffic Forecasting by secret19981209 is a document available to read on EtoBox.

The document presents a novel Decoupled Spatial-Temporal Framework (DSTF) for traffic forecasting, which separates diffusion and inherent traffic signals to improve prediction accuracy. It introduces an instantiation called Decoupled Dynamic Spatial-Temporal Graph Neural Network (D2 STGNN) that captures complex spatial-temporal correlations and incorporates a dynamic graph learning module. Extensive experiments demonstrate that this framework significantly outperforms existing methods in traffic forecasting

Author

secret19981209

Language

EN