net.sourceforge.cilib.nn.components.Neuron Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of cilib-library Show documentation
A library of composable components enabling simpler Computational Intelligence
There is a newer version: 0.8
/**           __  __
 *    _____ _/ /_/ /_    Computational Intelligence Library (CIlib)
 *   / ___/ / / / __ \   (c) CIRG @ UP
 *  / /__/ / / / /_/ /   http://cilib.net
 *  \___/_/_/_/_.___/
 */
package net.sourceforge.cilib.nn.components;

import net.sourceforge.cilib.functions.activation.ActivationFunction;
import net.sourceforge.cilib.functions.activation.Sigmoid;
import net.sourceforge.cilib.nn.architecture.NeuralInputSource;
import net.sourceforge.cilib.type.types.container.Vector;
import net.sourceforge.cilib.util.Cloneable;

/**
 * Class represents a neuron within a neural network and encapsulates weights
 * (that lead in to the neuron), an activation function and also stores its
 * current activation.
 */
public class Neuron implements Cloneable {

    private double activation; // the stored activation after the last calculateActivation call.
    private ActivationFunction activationFunction; // the neuron's activation function.
    private Vector weights; // the set of input weights for the neuron.

    /**
     * Default constructor.
     * Activation is set to 0.0. The default Activation Function is {@link Sigmoid}
     * and the weights are initialised as an empty {@link Vector}.
     */
    public Neuron() {
        activation = 0.0;
        activationFunction = new Sigmoid();
        weights = Vector.of();
    }

    /**
     * Copy constructor
     * @param neuron the neuron to copy.
     */
    public Neuron(Neuron neuron) {
        activation = neuron.getActivation();
        activationFunction = (ActivationFunction) neuron.getActivationFunction().getClone();
        weights = Vector.copyOf(neuron.weights);
    }

    /**
     * {@inheritDoc }
     */
    @Override
    public Object getClone() {
        return new Neuron(this);
    }

    /**
     * Calculates the neuron's activation by taking the dot product of the
     * netInputSource and the weights Vector and passing the result to the
     * activation function.
     * @param netInputSource the neuron's input source.
     * @return the neuron's activation.
     */
    public double calculateActivation(NeuralInputSource netInputSource) {
        double netInput = 0.0;

        int size = this.weights.size();
        for (int i = 0; i < size; i++) {
            netInput += netInputSource.getNeuralInput(i) * this.weights.doubleValueOf(i);
        }
        this.activation = activationFunction.apply(netInput);
        return activation;
    }

    /**
     * Gets the number of weights leading in to the neuron.
     * @return the number of input weights.
     */
    public int getNumWeights() {
        return this.weights.size();
    }

    /**
     * Gets the stored activation.
     * @return the stored activation after the last call of {@link #calculateActivation(net.sourceforge.cilib.nn.architecture.NeuralInputSource). }
     */
    public double getActivation() {
        return activation;
    }

    /**
     * Sets the neuron's activation.
     * @param activation the neuron's new activation.
     */
    public void setActivation(double activation) {
        this.activation = activation;
    }

    /**
     * Gets the neuron's activation function.
     * @return the neuron's activation function.
     */
    public ActivationFunction getActivationFunction() {
        return activationFunction;
    }

    /**
     * Sets the neuron's activation function.
     * @param activationFunction the neuron's activation function.
     */
    public void setActivationFunction(ActivationFunction activationFunction) {
        this.activationFunction = activationFunction;
    }

    /**
     * Gets the neuron's weights.
     * @return the neuron's weights.
     */
    public Vector getWeights() {
        return weights;
    }

    /**
     * Sets the neuron's weights.
     * @param weights the neuron's new weights.
     */
    public void setWeights(Vector weights) {
        this.weights = weights;
    }

    public boolean isBias() {
        return false;
    }
}