A Bayesian Neural Network-Based Method to Calibrate Microscopic Traffic Simulators

Table 4

Results of the calibration methods.

Method							MSE (optimal fitness value)	Average calculation time (seconds)
BNN-GA	3.75	−1.35	−2.70	0.38	2.32	−0.03	0.091	18.898
BNN-ES	5.00	−2.14	−2.21	0.22	1.74	2.51	0.018	16.467
BNN-BA	4.73	−2.25	−1.62	−1.10	−0.62	−0.40	0.001	8.054
ANN-GA	4.99	−0.58	−2.99	−0.27	−2.99	−2.99	0.199	30.298