deepnetts.net.layers.InputLayer Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of deepnetts-core Show documentation
Deep Learning Engine in pure Java. Base for reference implementation of JSR381
The newest version!
/**
 *  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;

import deepnetts.util.Tensor;

/**
 * Input layer in neural network, which accepts external input, and sends it to next layer in a network.
 * It is always the first layer in the network. Input can be 1D, 2D or 3D vectors/tensors of float values.
 *
 * @author Zoran Sevarac
 */
public class InputLayer extends AbstractLayer {

	
	private static final long serialVersionUID = 4852747073698145422L;
	
	
    /**
     * Creates input layer with specified width, height, and depth (number of
     * channels).
     *
     * @param width layer width
     * @param height layer height
     * @param depth layer depth (number of input channels)
     */
    public InputLayer(int width, int height, int depth) {
        this.width = width;
        this.height = height;
        this.depth = depth;  
        init();
    }

    /**
     * Creates input layer with specified width and height, with depth=1 (single channel).
     *
     * @param width layer width
     * @param height layer height
     */
    public InputLayer(int width, int height) {
        this.width = width;
        this.height = height;
        this.depth = 1; 
        init();
    }

    /**
     * Creates input layer with specified width, and with height and depth equals to one.
     *
     * @param width layer width
     */
    public InputLayer(int width) {
        this.width = width;
        this.height = 1;
        this.depth = 1;
        init();
    }

    /**
     * Initialize this layer in network.
     */
    @Override
    public final void init() {
        inputs = new Tensor(height, width, depth);
        outputs = inputs;  
    }

    /**
     * Sets network input
     *
     * @param in input matrix/array
     */
    public void setInput(Tensor in) {
        // TODO: check input tensor dimensions and throw exception if they dont match
        inputs.setValues(in.getValues());
    }

    /**
     * This method does nothing in input layer
     */
    @Override
    public void forward() {
        throw new IllegalStateException("This method does nothing and should never be called");
    }

    /**
     * This method does nothing in input layer
     */
    @Override
    public void backward() {
        throw new IllegalStateException("This method does nothing and should never be called");
    }

    /**
     * This method does nothing in input layer
     */
    @Override
    public void applyWeightChanges() {
    }
    
    @Override
    public String toString() {
        if (height==1 && depth==1)
            return "Input Layer { width:"+width+", height:"+height+", depth:"+depth+" }";
        else
            return "Input Layer { width:"+width+", height:"+height+", depth:"+depth+" }";
    }

}