CompositeNeuralNetworks.java

// Composite design pattern - neural networks

package structural.composite;

import java.util.ArrayList;
import java.util.Collections;
import java.util.Iterator;
import java.util.Spliterator;
import java.util.function.Consumer;

interface SomeNeurons extends Iterable<Neuron>
{
    default void connectTo(SomeNeurons other)
    {
        if (this == other) return;

        for (Neuron from : this)
            for (Neuron to : other)
            {
                from.out.add(to);
                to.in.add(from);
            }
    }
}

class Neuron implements SomeNeurons
{
    public ArrayList<Neuron> in, out;

    @Override
    public Iterator<Neuron> iterator() {
        return Collections.singleton(this).iterator();
    }

    @Override
    public void forEach(Consumer<? super Neuron> action) {
        action.accept(this);
    }

    @Override
    public Spliterator<Neuron> spliterator() {
        return Collections.singleton(this).spliterator();
    }

    // Connecting a pair of neurons
//    public void connectTo(Neuron other)
//    {
//        out.add(other);
//        other.in.add(this);
//    }
}

// ArrayList already implements Iterable<Neuron>
class NeuronLayer extends ArrayList<Neuron>
    implements SomeNeurons
{

}

class CompositeNetworksDemo
{
    public static void main(String[] args) {
        Neuron neuron = new Neuron();
        Neuron neuron2 = new Neuron();
        NeuronLayer layer = new NeuronLayer();
        NeuronLayer layer2 = new NeuronLayer();

        neuron.connectTo(neuron2);
        neuron.connectTo(layer);
        layer.connectTo(neuron);
        layer.connectTo(layer2);
    }
}