Lane-Changing Behavior Prediction Based on Game Theory and Deep Learning

Table 3

Prediction accuracy of different algorithm models.

Algorithm model	Number of hidden layers	Number of parameters	Prediction accuracy with game theory (%)	Prediction accuracy without game theory (%)
RBFNN	1	350	86.31	69.55
BPNN	2	63	84.89	75.36
CNN	2	48003	89.35	82.40
LSTM	1	167200	92.17	83.45
CNN-LSTM	2 + 1	90048	92.65	82.12
LSTM-CNN	1 + 1	167803	94.56	86.17