Hidden layer 1

The number of input neurons is equal to the output of the input layer. Then the number of outputs is an arbitrary value. We set a smaller value considering very few inputs and features:

DenseLayer hidden_layer_1 = new DenseLayer.Builder()
                .weightInit(WeightInit.XAVIER)
                .activation(Activation.RELU)
                .nIn(16).nOut(32)
                .build();