Table 4.

Performance of the neural fitting algorithms in conductivity estimation.

Algorithm name	Network size	Results	RMSE	R2
Levenberg–Marquardt	10	Training	0.42	0.94
Levenberg–Marquardt	10	Validation	0.49	0.91
Levenberg–Marquardt	10	Testing	0.53	0.88
Levenberg–Marquardt	100	Training	0.24	0.98
Levenberg–Marquardt	100	Validation	0.25	0.94
Levenberg–Marquardt	100	Testing	0.4	0.96
Bayesian regularization	50	Training	0.14	0.99
Bayesian regularization	50	Testing	0.18	0.99
Scaled conjugate gradient	100	Training	0.55	0.90
Scaled conjugate gradient	100	Validation	0.50	0.88
Scaled conjugate gradient	100	Testing	0.56	0.88