Table 3 Performance statistics for AIS trajectory prediction

From: Building adaptive knowledge bases for evolving continual learning models

Methods

ADE (↓)

FDE (↓)

IAcc (↑)

Linear

0.41 ± 0.04

1.04 ± 0.14

4.59% ± 1.01

Bi-GRU30

0.39 ± 0.12

0.92 ± 0.22

49.52% ± 5.10

Bi-LSTM31

0.34 ± 0.09

0.74 ± 0.21

68.16% ± 6.91

Traj-MAE33

0.33 ± 0.14

0.71 ± 0.25

70.26% ± 6.51

TrAISformer32

0.33 ± 0.12

0.72 ± 0.21

71.41% ± 5.94

PromptCast34

0.38 ± 0.11

0.79 ± 0.31

62.82% ± 6.11

PromptCast+CABLE

0.31 ± 0.09

0.70 ± 0.30

72.22% ± 6.09

  1. As indicated by arrows, it is desired to have a lower ADE and FDE, and a higher IAcc.
Back to article page