org.deeplearning4j.example.mnist.DenoisingAutoEncoderMnistExample Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of deeplearning4j-examples Show documentation

Examples of training different data sets

The newest version!

package org.deeplearning4j.example.mnist;

import org.apache.commons.math3.random.MersenneTwister;
import org.deeplearning4j.da.DenoisingAutoEncoder;
import org.deeplearning4j.datasets.DataSet;
import org.deeplearning4j.datasets.iterator.DataSetIterator;
import org.deeplearning4j.datasets.iterator.impl.MnistDataSetIterator;
import org.deeplearning4j.datasets.iterator.impl.RawMnistDataSetIterator;
import org.deeplearning4j.datasets.mnist.draw.DrawMnistGreyScale;
import org.deeplearning4j.nn.NeuralNetwork.LossFunction;
import org.deeplearning4j.nn.NeuralNetwork.OptimizationAlgorithm;
import org.deeplearning4j.plot.FilterRenderer;
import org.deeplearning4j.util.MatrixUtil;
import org.jblas.DoubleMatrix;

public class DenoisingAutoEncoderMnistExample {

	/**
	 * @param args
	 */
	public static void main(String[] args) throws Exception {
		DenoisingAutoEncoder autoEncoder = new DenoisingAutoEncoder.Builder()
		.numberOfVisible(784).numHidden(500).normalizeByInputRows(true).withLossFunction(LossFunction.NEGATIVELOGLIKELIHOOD)
		.useAdaGrad(true).useRegularization(true).withSparsity(0).withL2(0.01)
		.withOptmizationAlgo(OptimizationAlgorithm.GRADIENT_DESCENT)
		.withMomentum(0.5).build();


		//batches of 10, 60000 examples total
		DataSetIterator iter = new RawMnistDataSetIterator(10,30);
		for(int i = 0;i < 20; i++) {
			while(iter.hasNext()) {
				DataSet next = iter.next();
				//train with k = 1 0.01 learning rate and 1000 epochs
				autoEncoder.trainTillConvergence(next.getFirst(), 1e-1, new Object[]{0.6,1e-1,1000});


			}


			iter.reset();

		}

		FilterRenderer render = new FilterRenderer();
		render.renderFilters(autoEncoder.getW(), "example-render.jpg", 28, 28);




		//Iterate over the data set after done training and show the 2 side by side (you have to drag the test image over to the right)
		while(iter.hasNext()) {
			DataSet first = iter.next();
			DoubleMatrix reconstruct = autoEncoder.reconstruct(first.getFirst());
			for(int j = 0; j < first.numExamples(); j++) {

				DoubleMatrix draw1 = first.get(j).getFirst().mul(255);
				DoubleMatrix reconstructed2 = reconstruct.getRow(j);
				DoubleMatrix draw2 = MatrixUtil.binomial(reconstructed2,1,new MersenneTwister(123)).mul(255);

				DrawMnistGreyScale d = new DrawMnistGreyScale(draw1);
				d.title = "REAL";
				d.draw();
				DrawMnistGreyScale d2 = new DrawMnistGreyScale(draw2,1000,1000);
				d2.title = "TEST";
				d2.draw();
				Thread.sleep(10000);
				d.frame.dispose();
				d2.frame.dispose();
			}


		}

	}

}