AI Predicts Traffic: A Spatio-Temporal Graph Neural Network for Network-Wide Traffic Volume Estimation

This research explores a novel application of graph neural networks in traffic management, specifically estimating traffic volume using speed profiles. The use of a directed spatial attention mechanism suggests an attempt to capture complex spatial dependencies within traffic networks.