Table 8 Temporal graph attention network (TGAT) output for key feature indicators across time.

From: Design of an integrated model with temporal graph attention and transformer-augmented RNNs for enhanced anomaly detection

Time (s)

Node (Camera)

Attention weight

Long-term dependency score

Anomaly indicator

0–10

Camera 1

0.45

0.67

Normal

10–20

Camera 2

0.50

0.72

Normal

20–30

Camera 3

0.75

0.88

Anomalous

30–40

Camera 4

0.35

0.55

Normal

40–50

Camera 5

0.60

0.80

Normal

Back to article page