Peak Ground Acceleration Prediction by Artificial Neural Networks for Northwestern Turkey

Table 3

The final architectures of the models (i), (ii), (iii), (iv).

Model	Inputs	Outputs	FFBP method			GRNN method	RBF method
			ℎ𝑚	𝑓(⋅)	𝑓(⋅)	Spread values	Spread values
(i)	𝑀𝑤, FD, HD and SC	PGA for E-W	5	tan-sigmoid	log-sigmoid	0.015	0.022
(ii)	𝑀𝑤, FD, HD and SC	PGA for N-S	5	log-sigmoid	log-sigmoid	0.062	0.039
(iii)	𝑀𝑤, FD, HD and SC	PGA for U-D	2	tan-sigmoid	log-sigmoid	0.076	0.018
(iv)	𝑀𝑤, FD, HD, SC and D	Max. PGA	7	log-sigmoid	log-sigmoid	0.28	0.060