Table 5 Full information about the developed ANN models.

From: Experimental-based statistical models for the tensile characterization of synthetic fiber ropes: a machine learning approach

Network Architecture

Layers

Input

Hidden

Output

Layers No

1-layer

1-layer

1-layer

Neurons No

11-Neurons

1 to 30 Neurons (It will be optimized)

3-Neuron

Connection pattern

 

Multilayer Feed-Forward Network

 

Activation functions

Identity function \(\psi (u)\, = \,u\)

Hyperbolic Tangent Sigmoid function \(\psi (u) = \frac{{e^{u} - e^{{ - u}} }}{{e^{u} + e^{{ - u}} }}\)

Identity function \(\psi (u)\, = \,u\)

Training algorithm

Levenberg–Marquardt Backpropagation

Data splitting

fivefold cross validation (70% training + 15% Validation + 15% Testing)

Cost function

Root Mean Squared Error (RMSE)

Addressing nominal variables

Dummy parameters are generated for the nominal input variables

Back to article page