Table 8 Model hyper-parameters

Multi-layer perceptron network (MLP)

[55, 57]

Number of hidden layers: 3

Optimization method: scaled conjugate gradient backpropagation, learning rate self-adapting

Neurons per hidden layer: 50, 25, 10

Loss function: mean-squared error

Radial-basis function network with relevance (rRBF)

[58,59,60]

Number of radial basis functions: 30

Optimization method: scaled non-linear conjugate gradient, learning rate self adapting

Loss function: mean-squared error

Partially least square (PLS)

[61]

Number of components: 20

Linear discriminance model (LDA)

[62]

No hyper-parameters