Automatic Traffic Anomaly Detection on the Road Network with Spatial-Temporal Graph Neural Network Representation Learning

Table 2

Notations used in this paper.

Symbol	Description
	Data of different stations at the moment of
	The station’s embedding vector
	Adjacency matrix donates the relationship between node and node
	The candidate relation node set of
	The similarity between the embedding vector of the node and the embedding vector of its candidate node
	The score of the nodes through the graph attention mechanism
	The attention coefficient
	Error value of the site at the time
	A smoothed score will be used to decide on an exception has happened or not