Prediction of Concrete Compressive Strength by Evolutionary Artificial Neural Networks

Table 3

Optimal structure of MLP, FF, RBF, and TLRN models obtained from GA.

GA for all models		Learning algorithm	Transfer function	Number of hidden neuron	Number of hidden layer	Number of output	Number of input	Model	Number
Crossover	One point	Momentum	TanhAxon	4	1	1	7	MLP	1
Crossover probability	0.9	Step	LinearAxon	15	1			FF	2
Mutation probability	0.01	Delta Bar Delta	LinearSigmoidAxon	4_4	2			RBF	4
Generation	100	Momentum	TanhAxon	5_10	2			TLRN	3