deepnetts.net.layers.activation.ActivationFunction Maven / Gradle / Ivy

Go to download
/**
 *  DeepNetts is pure Java Deep Learning Library with support for Backpropagation
 *  based learning and image recognition.
 *
 *  Copyright (C) 2017  Zoran Sevarac 
 *
 * This file is part of DeepNetts.
 *
 * DeepNetts is free software: you can redistribute it and/or modify it under
 * the terms of the GNU General Public License as published by the Free Software
 * Foundation, either version 3 of the License, or (at your option) any later
 * version.
 *
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
 * Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program. If not, see .package
 * deepnetts.core;
 */

package deepnetts.net.layers.activation;

import deepnetts.util.DeepNettsException;

/**
 * Common base interface for all activation functions used in layers.
 * Classes implementing this interface should provide methods for calculating
 * value and first derivative of the activation function.
 * Activation function performs non-linear transformation of its input
 * before its sent to layer output.
 * First derivative of a function shows how fast and in what direction function
 * is changing if its input changes, and it is used by training algorithm.
 *
 * For more see https://en.wikipedia.org/wiki/Activation_function
 *
 * @see ActivationType
 * @author Zoran Sevarac
 */
public interface ActivationFunction {

    /**
     * Returns the value of activation function for specified input x
     * @param x input for activation
     * @return value of activation function
     */
    public float getValue(float x);


    /**
     * Returns the first derivative of activation function for specified output y
     * @param y output of activation function
     * @return first derivative of activation function
     */
    public float getPrime(float y);


    /**
     * Creates and returns specified type of activation function.
     * A factory method for creating activation functions;
     *
     * @param type type of the activation function
     *
     * @return returns instance of specified activation function type
     */
    public static ActivationFunction create(ActivationType type) {
        switch (type) {
            case LINEAR:
                return new Linear();
            case RELU:
                return new Relu();
            case SIGMOID:
                return new Sigmoid();
            case TANH:
                return new Tanh();
            default:
                throw new DeepNettsException("Unknown activation function:" + type);
        }
    }

}