Evolution of Deep Neural Network Architecture Using Particle Swarm Optimization to Improve the Performance in Determining the Friction Angle of Soil

Table 8

Summary of the 300 simulations using MAE criteria.

Model	Data set	Average (^o)	Min (^o)	Max (^o)	SD
10-HD-R^2	Training	1.188	0.883	1.423	0.0899
	Validation	1.721	1.392	2.509	0.2345
	Testing	1.340	1.036	1.715	0.1339
10-HD-MAE	Training	1.636	1.050	2.560	0.3031
	Validation	2.105	1.473	3.455	0.3411
	Testing	1.823	1.243	2.870	0.271
8-HD-RMSE	Training	1.268	1.101	2.107	0.1534
	Validation	1.706	1.267	3.444	0.2564
	Testing	1.523	1.238	2.338	0.1923