Table 6 Hyperparameters for custom ANN models

Hidden layer count

{1, 2, 3, 4, 5, 6, 7}

Node count per hidden layer

{16, 32, 64, 128, 256}

Dropout layer

{true, false}

Regularizer

{None, L1, L2}

Regularization term

{0.1, 0.01, 0.001}

Learning rate schedule

{constant,  linear,  staircase exponential,  continuous exponential}

Initial learning rate

{0.1, 0.01, 0.001}

Learning rate decay rate

{0.1, 0.25, 0.5}

Optimizer

{SGD, Adam}

Optimizer momentum

{0.8, 0.9, 0.99}