import com.neuralnetwork.Backpropagation;
import com.neuralnetwork.Layer;
import java.util.Random;

/************************
 * @author Max Rafiandy *
 ************************/
public class Main {

    public static void main(String[] args) {
        final Random random = new Random();
        double[][] learning_x = { {0, 0}, {0, 1}, {0, 1}, {1, 0} };
        double[][] target_out = { {0}, {0}, {0}, {1} };

        Layer input_layer = new Layer(learning_x[0].length, Layer.INPUT);
        Layer hidden_layer = new Layer(learning_x[0].length
                + target_out[0].length, Layer.HIDDEN);
        Layer output_layer = new Layer(target_out[0].length, Layer.OUTPUT);

        for(int j=0; j<hidden_layer.neuron.length; j++) {
            hidden_layer.bias[j] = random.nextDouble();
            hidden_layer.weight = new double[input_layer.neuron.length]
                    [hidden_layer.neuron.length];
            for (int i=0; i<input_layer.neuron.length; i++) {
                hidden_layer.weight[i][j] = random.nextDouble();
            }
        }

        for(int k=0; k<output_layer.neuron.length; k++) {
            output_layer.bias[k] = random.nextDouble();
            output_layer.weight = new double[hidden_layer.neuron.length]
                    [output_layer.neuron.length];
            for (int j=0; j<hidden_layer.neuron.length; j++) {
                output_layer.weight[j][k] = random.nextDouble();
            }
        }

        while (true) {
            double mse = 0;
            for (int i=0; i<learning_x.length; i++) {
                input_layer.neuron = learning_x[i];
                mse += Backpropagation.learning(input_layer, hidden_layer,
                        output_layer, target_out[i], 0.5);
            }
            System.out.println("Updating Error: " + 0.5*mse);
            if (0.5 * mse < 1e-5) break;
        }
    }
}