Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance.
Project price only 1 $
You can buy this project and download/modify it how often you want.
org.nd4j.linalg.lossfunctions.LossFunctionTests Maven / Gradle / Ivy
package org.nd4j.linalg.lossfunctions;
import org.nd4j.linalg.api.activation.Activations;
import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.factory.Nd4j;
import org.junit.Test;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import static org.junit.Assert.assertEquals;
/**
* Testing loss function
*
* @author Adam Gibson
*/
public abstract class LossFunctionTests {
private static Logger log = LoggerFactory.getLogger(LossFunctionTests.class);
@Test
public void testReconEntropy() {
Nd4j.factory().setOrder('f');
INDArray input = Nd4j.create(
new double[]{1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0}, new int[]{7, 6});
INDArray w = Nd4j.create(
new double[]{-0.005221025740321007, -0.0025006434506737304, -0.013585431005440437, -0.021996946700655134, 0.007678447599654643, -0.0037941287958231052, -0.014933056402715545, -0.012875289265542541, 0.001635482018910717, 0.00893829162129914, 0.017003519496588012, -0.004271078749979736, 0.0015816435136811352, 0.008638074705740708, -0.004393004605647038, -0.006249587919004255, -0.011017655538216209, -0.0015862988109404338, 0.01079760516931169, -0.0010491291520692704, 0.006626023289526534, 0.004658989751677583, -0.0022132443508813535, -0.00979834812384658}, new int[]{6, 4});
INDArray vBias = Nd4j.create(
new double[]{0.0, 0.0, 0.0, 0.0, 0.0, 0.0}, new int[]{6});
INDArray hBias = Nd4j.create(
new double[]{0.0, 0.0, 0.0, 0.0}, new int[]{4});
double reconEntropy = LossFunctions.reconEntropy(input,hBias,vBias,w, Activations.sigmoid());
double assertion = -0.5937198421625942;
assertEquals(assertion,reconEntropy,1e-1);
}
@Test
public void testRMseXent() {
INDArray in = Nd4j.create(new float[][]{{1, 2}, {3, 4}});
INDArray out = Nd4j.create(new float[][]{{5, 6}, {7, 8}});
double diff = LossFunctions.score(in, LossFunctions.LossFunction.RMSE_XENT,out,0,false);
assertEquals(8,diff,1e-1);
}
@Test
public void testMcXent() {
INDArray in = Nd4j.create(new float[][]{{1, 2}, {3, 4}});
INDArray out = Nd4j.create(new float[][]{{5, 6}, {7, 8}});
LossFunctions.score(in, LossFunctions.LossFunction.MCXENT,out,0,false);
}
@Test
public void testNegativeLogLikelihood() {
INDArray softmax = Nd4j.create(new double[][]{{0.6,0.4},{0.7,0.3}});
INDArray trueLabels = Nd4j.create(new double[][]{{1,0},{0,1}});
double score = LossFunctions.score(trueLabels, LossFunctions.LossFunction.NEGATIVELOGLIKELIHOOD,softmax,0,false);
assertEquals(1.71479842809,score,1e-1);
INDArray softmax2 = Nd4j.create(new double[][]{{0.33,0.33,0.33},{0.33,0.33,0.33}});
INDArray trueLabels2 = Nd4j.create(new double[][]{{1,0,0},{1,0,0}});
double score2 = LossFunctions.score(trueLabels2, LossFunctions.LossFunction.NEGATIVELOGLIKELIHOOD,softmax2,0,false);
assertEquals(1.90961775772,score2,1e-1);
}
}
