Java Code Examples for org.nd4j.linalg.dataset.api.preprocessor.NormalizerStandardize#fit()

@Test
public void testRevert() {
    double tolerancePerc = 0.01; // 0.01% of correct value
    int nSamples = 500;
    int nFeatures = 3;

    INDArray featureSet = Nd4j.randn(nSamples, nFeatures);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleDataSet);
    DataSet transformed = sampleDataSet.copy();
    myNormalizer.transform(transformed);
    //System.out.println(transformed.getFeatures());
    myNormalizer.revert(transformed);
    //System.out.println(transformed.getFeatures());
    INDArray delta = Transforms.abs(transformed.getFeatures().sub(sampleDataSet.getFeatures()))
                    .div(sampleDataSet.getFeatures());
    double maxdeltaPerc = delta.max(0, 1).mul(100).getDouble(0, 0);
    assertTrue(maxdeltaPerc < tolerancePerc);
}
 

@Test
public void testBruteForce4d() {
    Construct4dDataSet imageDataSet = new Construct4dDataSet(10, 5, 10, 15);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(imageDataSet.sampleDataSet);
    assertEquals(imageDataSet.expectedMean, myNormalizer.getMean());

    float aat = Transforms.abs(myNormalizer.getStd().div(imageDataSet.expectedStd).sub(1)).maxNumber().floatValue();
    float abt = myNormalizer.getStd().maxNumber().floatValue();
    float act = imageDataSet.expectedStd.maxNumber().floatValue();
    System.out.println("ValA: " + aat);
    System.out.println("ValB: " + abt);
    System.out.println("ValC: " + act);
    assertTrue(aat < 0.05);

    NormalizerMinMaxScaler myMinMaxScaler = new NormalizerMinMaxScaler();
    myMinMaxScaler.fit(imageDataSet.sampleDataSet);
    assertEquals(imageDataSet.expectedMin, myMinMaxScaler.getMin());
    assertEquals(imageDataSet.expectedMax, myMinMaxScaler.getMax());

    DataSet copyDataSet = imageDataSet.sampleDataSet.copy();
    myNormalizer.transform(copyDataSet);
}
 

public static INDArray generateOutput(File inputFile, String modelFilePath) throws IOException, InterruptedException {
    final File modelFile = new File(modelFilePath);
    final MultiLayerNetwork network = ModelSerializer.restoreMultiLayerNetwork(modelFile);
    final RecordReader recordReader = generateReader(inputFile);
    //final INDArray array = RecordConverter.toArray(recordReader.next());
    final NormalizerStandardize normalizerStandardize = ModelSerializer.restoreNormalizerFromFile(modelFile);
    //normalizerStandardize.transform(array);
    final DataSetIterator dataSetIterator = new RecordReaderDataSetIterator.Builder(recordReader,1).build();
    normalizerStandardize.fit(dataSetIterator);
    dataSetIterator.setPreProcessor(normalizerStandardize);
    return network.output(dataSetIterator);

}
 

@Test
public void testUnderOverflow() {
    // This dataset will be basically constant with a small std deviation
    // And the constant is large. Checking if algorithm can handle
    double tolerancePerc = 1; //Within 1 %
    double toleranceAbs = 0.0005;
    int nSamples = 1000;
    int bSize = 10;
    int x = -1000000, y = 1000000;
    double z = 1000000;

    INDArray featureX = Nd4j.rand(nSamples, 1).mul(1).add(x);
    INDArray featureY = Nd4j.rand(nSamples, 1).mul(2).add(y);
    INDArray featureZ = Nd4j.rand(nSamples, 1).mul(3).add(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);

    INDArray theoreticalMean = Nd4j.create(new float[] {x, y, (float) z}).castTo(Nd4j.defaultFloatingPointType()).reshape(1, -1);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleIter);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray meanDeltaPerc = meanDelta.mul(100).div(theoreticalMean);
    assertTrue(meanDeltaPerc.max(1).getDouble(0) < tolerancePerc);

    //this just has to not barf
    //myNormalizer.transform(sampleIter);
    myNormalizer.transform(sampleDataSet);
}
 

@Override
public void normalize() {
    //FeatureUtil.normalizeMatrix(getFeatures());
    NormalizerStandardize inClassPreProcessor = new NormalizerStandardize();
    inClassPreProcessor.fit(this);
    inClassPreProcessor.transform(this);
}
 

@Test
public void testConstant() {
    double tolerancePerc = 10.0; // 10% of correct value
    int nSamples = 500;
    int nFeatures = 3;
    int constant = 100;

    INDArray featureSet = Nd4j.zeros(nSamples, nFeatures).add(constant);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);


    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleDataSet);
    //Checking if we gets nans
    assertFalse(Double.isNaN(myNormalizer.getStd().getDouble(0)));

    myNormalizer.transform(sampleDataSet);
    //Checking if we gets nans, because std dev is zero
    assertFalse(Double.isNaN(sampleDataSet.getFeatures().min(0, 1).getDouble(0)));
    //Checking to see if transformed values are close enough to zero
    assertEquals(Transforms.abs(sampleDataSet.getFeatures()).max(0, 1).getDouble(0, 0), 0,
                    constant * tolerancePerc / 100.0);

    myNormalizer.revert(sampleDataSet);
    //Checking if we gets nans, because std dev is zero
    assertFalse(Double.isNaN(sampleDataSet.getFeatures().min(0, 1).getDouble(0)));
    assertEquals(Transforms.abs(sampleDataSet.getFeatures().sub(featureSet)).min(0, 1).getDouble(0), 0,
                    constant * tolerancePerc / 100.0);
}
 

@Test
public void testRpTreeMaxNodes() throws Exception {
    DataSetIterator mnist = new MnistDataSetIterator(150,150);
    RPForest rpTree = new RPForest(4,4,"euclidean");
    DataSet d = mnist.next();
    NormalizerStandardize normalizerStandardize = new NormalizerStandardize();
    normalizerStandardize.fit(d);
    rpTree.fit(d.getFeatures());
    for(RPTree tree : rpTree.getTrees()) {
        for(RPNode node : tree.getLeaves()) {
            assertTrue(node.getIndices().size() <= rpTree.getMaxSize());
        }
    }

}
 

@Test
public void testUnderOverflow() {
    // This dataset will be basically constant with a small std deviation
    // And the constant is large. Checking if algorithm can handle
    double tolerancePerc = 1; //Within 1 %
    double toleranceAbs = 0.0005;
    int nSamples = 1000;
    int bSize = 10;
    int x = -1000000, y = 1000000;
    double z = 1000000;

    INDArray featureX = Nd4j.rand(nSamples, 1).mul(1).add(x);
    INDArray featureY = Nd4j.rand(nSamples, 1).mul(2).add(y);
    INDArray featureZ = Nd4j.rand(nSamples, 1).mul(3).add(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);

    INDArray theoreticalMean = Nd4j.create(new double[] {x, y, z});

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleIter);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray meanDeltaPerc = meanDelta.mul(100).div(theoreticalMean);
    assertTrue(meanDeltaPerc.max(1).getDouble(0, 0) < tolerancePerc);

    //this just has to not barf
    //myNormalizer.transform(sampleIter);
    myNormalizer.transform(sampleDataSet);
}
 

@Override
public void normalize() {
    //FeatureUtil.normalizeMatrix(getFeatures());
    NormalizerStandardize inClassPreProcessor = new NormalizerStandardize();
    inClassPreProcessor.fit(this);
    inClassPreProcessor.transform(this);
}
 

@Test
public void testBruteForce() {
    /* This test creates a dataset where feature values are multiples of consecutive natural numbers
       The obtained values are compared to the theoretical mean and std dev
     */
    double tolerancePerc = 0.01;
    int nSamples = 5120;
    int x = 1, y = 2, z = 3;

    INDArray featureX = Nd4j.linspace(1, nSamples, nSamples).reshape(nSamples, 1).mul(x);
    INDArray featureY = featureX.mul(y);
    INDArray featureZ = featureX.mul(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = featureSet.dup().getColumns(0);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);

    double meanNaturalNums = (nSamples + 1) / 2.0;
    INDArray theoreticalMean =
                    Nd4j.create(new double[] {meanNaturalNums * x, meanNaturalNums * y, meanNaturalNums * z}).reshape(1, -1).castTo(Nd4j.defaultFloatingPointType());
    INDArray theoreticallabelMean = theoreticalMean.dup().getColumns(0);
    double stdNaturalNums = Math.sqrt((nSamples * nSamples - 1) / 12.0);
    INDArray theoreticalStd =
                    Nd4j.create(new double[] {stdNaturalNums * x, stdNaturalNums * y, stdNaturalNums * z}).reshape(1, -1).castTo(Nd4j.defaultFloatingPointType());
    INDArray theoreticallabelStd = theoreticalStd.dup().getColumns(0);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fitLabel(true);
    myNormalizer.fit(sampleDataSet);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray labelDelta = Transforms.abs(theoreticallabelMean.sub(myNormalizer.getLabelMean()));
    INDArray meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    INDArray labelDeltaPerc = labelDelta.div(theoreticallabelMean).mul(100);
    double maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);
    assertTrue(labelDeltaPerc.max(1).getDouble(0) < tolerancePerc);

    INDArray stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    INDArray stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    INDArray stdlabelDeltaPerc =
                    Transforms.abs(theoreticallabelStd.sub(myNormalizer.getLabelStd())).div(theoreticallabelStd);
    double maxStdDeltaPerc = stdDeltaPerc.max(1).mul(100).getDouble(0);
    double maxlabelStdDeltaPerc = stdlabelDeltaPerc.max(1).getDouble(0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
    assertTrue(maxlabelStdDeltaPerc < tolerancePerc);


    // SAME TEST WITH THE ITERATOR
    int bSize = 10;
    tolerancePerc = 0.1; // 1% of correct value
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);
    myNormalizer.fit(sampleIter);

    meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    stdDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
}
 

@Test
public void testBruteForce() {
    /* This test creates a dataset where feature values are multiples of consecutive natural numbers
       The obtained values are compared to the theoretical mean and std dev
     */
    double tolerancePerc = 0.01; // 0.01% of correct value
    int nSamples = 5120;
    int x = 1, y = 2, z = 3;

    INDArray featureX = Nd4j.linspace(1, nSamples, nSamples, DataType.DOUBLE).reshape(nSamples, 1).mul(x);
    INDArray featureY = featureX.mul(y);
    INDArray featureZ = featureX.mul(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);

    double meanNaturalNums = (nSamples + 1) / 2.0;
    INDArray theoreticalMean =
                    Nd4j.create(new double[] {meanNaturalNums * x, meanNaturalNums * y, meanNaturalNums * z}).reshape(1, -1);
    double stdNaturalNums = Math.sqrt((nSamples * nSamples - 1) / 12.0);
    INDArray theoreticalStd =
                    Nd4j.create(new double[] {stdNaturalNums * x, stdNaturalNums * y, stdNaturalNums * z}).reshape(1, -1);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleDataSet);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    double maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    INDArray stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    INDArray stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    double maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);

    // SAME TEST WITH THE ITERATOR
    int bSize = 10;
    tolerancePerc = 0.1; // 0.1% of correct value
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);
    myNormalizer.fit(sampleIter);

    meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
}
 

@Test
public void normalizationTests() {
    List<List<Writable>> data = new ArrayList<>();
    Schema.Builder builder = new Schema.Builder();
    int numColumns = 6;
    for (int i = 0; i < numColumns; i++)
        builder.addColumnDouble(String.valueOf(i));

    for (int i = 0; i < 5; i++) {
        List<Writable> record = new ArrayList<>(numColumns);
        data.add(record);
        for (int j = 0; j < numColumns; j++) {
            record.add(new DoubleWritable(1.0));
        }

    }

    INDArray arr = RecordConverter.toMatrix(DataType.DOUBLE, data);

    Schema schema = builder.build();
    JavaRDD<List<Writable>> rdd = sc.parallelize(data);
    assertEquals(schema, DataFrames.fromStructType(DataFrames.fromSchema(schema)));
    assertEquals(rdd.collect(), DataFrames.toRecords(DataFrames.toDataFrame(schema, rdd)).getSecond().collect());

    Dataset<Row> dataFrame = DataFrames.toDataFrame(schema, rdd);
    dataFrame.show();
    Normalization.zeromeanUnitVariance(dataFrame).show();
    Normalization.normalize(dataFrame).show();

    //assert equivalent to the ndarray pre processing
    NormalizerStandardize standardScaler = new NormalizerStandardize();
    standardScaler.fit(new DataSet(arr.dup(), arr.dup()));
    INDArray standardScalered = arr.dup();
    standardScaler.transform(new DataSet(standardScalered, standardScalered));
    DataNormalization zeroToOne = new NormalizerMinMaxScaler();
    zeroToOne.fit(new DataSet(arr.dup(), arr.dup()));
    INDArray zeroToOnes = arr.dup();
    zeroToOne.transform(new DataSet(zeroToOnes, zeroToOnes));

    INDArray zeroMeanUnitVarianceDataFrame =
                    RecordConverter.toMatrix(DataType.DOUBLE, Normalization.zeromeanUnitVariance(schema, rdd).collect());
    INDArray zeroMeanUnitVarianceDataFrameZeroToOne =
                    RecordConverter.toMatrix(DataType.DOUBLE, Normalization.normalize(schema, rdd).collect());
    assertEquals(standardScalered, zeroMeanUnitVarianceDataFrame);
    assertTrue(zeroToOnes.equalsWithEps(zeroMeanUnitVarianceDataFrameZeroToOne, 1e-1));

}
 

@Test
public void testBruteForce() {
    /* This test creates a dataset where feature values are multiples of consecutive natural numbers
       The obtained values are compared to the theoretical mean and std dev
     */
    double tolerancePerc = 0.01; // 0.01% of correct value
    int nSamples = 5120;
    int x = 1, y = 2, z = 3;

    INDArray featureX = Nd4j.linspace(1, nSamples, nSamples).reshape(nSamples, 1).mul(x);
    INDArray featureY = featureX.mul(y);
    INDArray featureZ = featureX.mul(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = Nd4j.zeros(nSamples, 1);
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);

    double meanNaturalNums = (nSamples + 1) / 2.0;
    INDArray theoreticalMean =
                    Nd4j.create(new double[] {meanNaturalNums * x, meanNaturalNums * y, meanNaturalNums * z});
    double stdNaturalNums = Math.sqrt((nSamples * nSamples - 1) / 12.0);
    INDArray theoreticalStd =
                    Nd4j.create(new double[] {stdNaturalNums * x, stdNaturalNums * y, stdNaturalNums * z});

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fit(sampleDataSet);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    double maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    INDArray stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    INDArray stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    double maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);

    // SAME TEST WITH THE ITERATOR
    int bSize = 10;
    tolerancePerc = 0.1; // 0.1% of correct value
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);
    myNormalizer.fit(sampleIter);

    meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
}
 

@Test
public void testBruteForce3dMaskLabels() {

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fitLabel(true);
    NormalizerMinMaxScaler myMinMaxScaler = new NormalizerMinMaxScaler();
    myMinMaxScaler.fitLabel(true);

    //generating a dataset with consecutive numbers as feature values. Dataset also has masks
    int samples = 100;
    INDArray featureScale = Nd4j.create(new float[] {1, 2, 10}).reshape(3, 1);
    int timeStepsU = 5;
    Construct3dDataSet sampleU = new Construct3dDataSet(featureScale, timeStepsU, samples, 1);
    int timeStepsV = 3;
    Construct3dDataSet sampleV = new Construct3dDataSet(featureScale, timeStepsV, samples, sampleU.newOrigin);
    List<DataSet> dataSetList = new ArrayList<DataSet>();
    dataSetList.add(sampleU.sampleDataSet);
    dataSetList.add(sampleV.sampleDataSet);

    DataSet fullDataSetA = DataSet.merge(dataSetList);
    DataSet fullDataSetAA = fullDataSetA.copy();
    //This should be the same datasets as above without a mask
    Construct3dDataSet fullDataSetNoMask =
                    new Construct3dDataSet(featureScale, timeStepsU + timeStepsV, samples, 1);

    //preprocessors - label and feature values are the same
    myNormalizer.fit(fullDataSetA);
    assertEquals(myNormalizer.getMean(), fullDataSetNoMask.expectedMean);
    assertEquals(myNormalizer.getStd(), fullDataSetNoMask.expectedStd);
    assertEquals(myNormalizer.getLabelMean(), fullDataSetNoMask.expectedMean);
    assertEquals(myNormalizer.getLabelStd(), fullDataSetNoMask.expectedStd);

    myMinMaxScaler.fit(fullDataSetAA);
    assertEquals(myMinMaxScaler.getMin(), fullDataSetNoMask.expectedMin);
    assertEquals(myMinMaxScaler.getMax(), fullDataSetNoMask.expectedMax);
    assertEquals(myMinMaxScaler.getLabelMin(), fullDataSetNoMask.expectedMin);
    assertEquals(myMinMaxScaler.getLabelMax(), fullDataSetNoMask.expectedMax);


    //Same Test with an Iterator, values should be close for std, exact for everything else
    DataSetIterator sampleIterA = new TestDataSetIterator(fullDataSetA, 5);
    DataSetIterator sampleIterB = new TestDataSetIterator(fullDataSetAA, 5);

    myNormalizer.fit(sampleIterA);
    assertEquals(myNormalizer.getMean(), fullDataSetNoMask.expectedMean);
    assertEquals(myNormalizer.getLabelMean(), fullDataSetNoMask.expectedMean);
    assertTrue(Transforms.abs(myNormalizer.getStd().div(fullDataSetNoMask.expectedStd).sub(1)).maxNumber()
                    .floatValue() < 0.01);
    assertTrue(Transforms.abs(myNormalizer.getLabelStd().div(fullDataSetNoMask.expectedStd).sub(1)).maxNumber()
                    .floatValue() < 0.01);

    myMinMaxScaler.fit(sampleIterB);
    assertEquals(myMinMaxScaler.getMin(), fullDataSetNoMask.expectedMin);
    assertEquals(myMinMaxScaler.getMax(), fullDataSetNoMask.expectedMax);
    assertEquals(myMinMaxScaler.getLabelMin(), fullDataSetNoMask.expectedMin);
    assertEquals(myMinMaxScaler.getLabelMax(), fullDataSetNoMask.expectedMax);
}
 

@Test
public void testRestoreUnsavedNormalizerFromInputStream() throws Exception {
    DataSet dataSet = trivialDataSet();

    NormalizerStandardize norm = new NormalizerStandardize();
    norm.fit(dataSet);

    ComputationGraph cg = simpleComputationGraph();
    cg.init();

    File tempFile = tempDir.newFile();
    ModelSerializer.writeModel(cg, tempFile, true);

    FileInputStream fis = new FileInputStream(tempFile);

    NormalizerStandardize restored = ModelSerializer.restoreNormalizerFromInputStream(fis);

    assertEquals(null, restored);
}
 

@Test
    public void testAutoEncoder() {
        //As above (testGradientMLP2LayerIrisSimple()) but with L2, L1, and both L2/L1 applied
        //Need to run gradient through updater, so that L2 can be applied

        Activation[] activFns = {Activation.SIGMOID, Activation.TANH};
        boolean[] characteristic = {false, true}; //If true: run some backprop steps first

        LossFunction[] lossFunctions = {LossFunction.MCXENT, LossFunction.MSE};
        Activation[] outputActivations = {Activation.SOFTMAX, Activation.TANH};

        DataNormalization scaler = new NormalizerMinMaxScaler();
        DataSetIterator iter = new IrisDataSetIterator(150, 150);
        scaler.fit(iter);
        iter.setPreProcessor(scaler);
        DataSet ds = iter.next();
        INDArray input = ds.getFeatures();
        INDArray labels = ds.getLabels();

        NormalizerStandardize norm = new NormalizerStandardize();
        norm.fit(ds);
        norm.transform(ds);

        double[] l2vals = {0.2, 0.0, 0.2};
        double[] l1vals = {0.0, 0.3, 0.3}; //i.e., use l2vals[i] with l1vals[i]

        for (Activation afn : activFns) {
            for (boolean doLearningFirst : characteristic) {
                for (int i = 0; i < lossFunctions.length; i++) {
                    for (int k = 0; k < l2vals.length; k++) {
                        LossFunction lf = lossFunctions[i];
                        Activation outputActivation = outputActivations[i];
                        double l2 = l2vals[k];
                        double l1 = l1vals[k];

                        Nd4j.getRandom().setSeed(12345);
                        MultiLayerConfiguration conf =
                                        new NeuralNetConfiguration.Builder()
                                                        .dataType(DataType.DOUBLE)
                                                        .updater(new NoOp())
                                                        .l2(l2).l1(l1)
                                                        .optimizationAlgo(OptimizationAlgorithm.CONJUGATE_GRADIENT)
                                                        .seed(12345L)
                                                        .dist(new NormalDistribution(0, 1))
                                                        .list().layer(0,
                                                                        new AutoEncoder.Builder().nIn(4).nOut(3)
                                                                                        .activation(afn).build())
                                                        .layer(1, new OutputLayer.Builder(lf).nIn(3).nOut(3)
                                                                        .activation(outputActivation).build())
                                                        .build();

                        MultiLayerNetwork mln = new MultiLayerNetwork(conf);
                        mln.init();

                        String msg;
                        if (doLearningFirst) {
                            //Run a number of iterations of learning
                            mln.setInput(ds.getFeatures());
                            mln.setLabels(ds.getLabels());
                            mln.computeGradientAndScore();
                            double scoreBefore = mln.score();
                            for (int j = 0; j < 10; j++)
                                mln.fit(ds);
                            mln.computeGradientAndScore();
                            double scoreAfter = mln.score();
                            //Can't test in 'characteristic mode of operation' if not learning
                            msg = "testGradMLP2LayerIrisSimple() - score did not (sufficiently) decrease during learning - activationFn="
                                            + afn + ", lossFn=" + lf + ", outputActivation=" + outputActivation
                                            + ", doLearningFirst=" + doLearningFirst + ", l2=" + l2 + ", l1=" + l1
                                            + " (before=" + scoreBefore + ", scoreAfter=" + scoreAfter + ")";
                            assertTrue(msg, scoreAfter < scoreBefore);
                        }

                        msg = "testGradMLP2LayerIrisSimple() - activationFn=" + afn + ", lossFn=" + lf
                                        + ", outputActivation=" + outputActivation + ", doLearningFirst="
                                        + doLearningFirst + ", l2=" + l2 + ", l1=" + l1;
                        if (PRINT_RESULTS) {
                            System.out.println(msg);
//                            for (int j = 0; j < mln.getnLayers(); j++)
//                                System.out.println("Layer " + j + " # params: " + mln.getLayer(j).numParams());
                        }

                        boolean gradOK = GradientCheckUtil.checkGradients(mln, DEFAULT_EPS, DEFAULT_MAX_REL_ERROR,
                                        DEFAULT_MIN_ABS_ERROR, PRINT_RESULTS, RETURN_ON_FIRST_FAILURE, input, labels);
                        assertTrue(msg, gradOK);
                        TestUtils.testModelSerialization(mln);
                    }
                }
            }
        }
    }
 

@Test
public void testBruteForce() {
    /* This test creates a dataset where feature values are multiples of consecutive natural numbers
       The obtained values are compared to the theoretical mean and std dev
     */
    double tolerancePerc = 0.01;
    int nSamples = 5120;
    int x = 1, y = 2, z = 3;

    INDArray featureX = Nd4j.linspace(1, nSamples, nSamples).reshape(nSamples, 1).mul(x);
    INDArray featureY = featureX.mul(y);
    INDArray featureZ = featureX.mul(z);
    INDArray featureSet = Nd4j.concat(1, featureX, featureY, featureZ);
    INDArray labelSet = featureSet.dup().getColumns(new int[] {0});
    DataSet sampleDataSet = new DataSet(featureSet, labelSet);

    double meanNaturalNums = (nSamples + 1) / 2.0;
    INDArray theoreticalMean =
                    Nd4j.create(new double[] {meanNaturalNums * x, meanNaturalNums * y, meanNaturalNums * z});
    INDArray theoreticallabelMean = theoreticalMean.dup().getColumns(new int[] {0});
    double stdNaturalNums = Math.sqrt((nSamples * nSamples - 1) / 12.0);
    INDArray theoreticalStd =
                    Nd4j.create(new double[] {stdNaturalNums * x, stdNaturalNums * y, stdNaturalNums * z});
    INDArray theoreticallabelStd = theoreticalStd.dup().getColumns(new int[] {0});

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fitLabel(true);
    myNormalizer.fit(sampleDataSet);

    INDArray meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    INDArray labelDelta = Transforms.abs(theoreticallabelMean.sub(myNormalizer.getLabelMean()));
    INDArray meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    INDArray labelDeltaPerc = labelDelta.div(theoreticallabelMean).mul(100);
    double maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);
    assertTrue(labelDeltaPerc.max(1).getDouble(0, 0) < tolerancePerc);

    INDArray stdDelta = Transforms.abs(theoreticalStd.sub(myNormalizer.getStd()));
    INDArray stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    INDArray stdlabelDeltaPerc =
                    Transforms.abs(theoreticallabelStd.sub(myNormalizer.getLabelStd())).div(theoreticallabelStd);
    double maxStdDeltaPerc = stdDeltaPerc.max(1).mul(100).getDouble(0, 0);
    double maxlabelStdDeltaPerc = stdlabelDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
    assertTrue(maxlabelStdDeltaPerc < tolerancePerc);


    // SAME TEST WITH THE ITERATOR
    int bSize = 10;
    tolerancePerc = 0.1; // 1% of correct value
    DataSetIterator sampleIter = new TestDataSetIterator(sampleDataSet, bSize);
    myNormalizer.fit(sampleIter);

    meanDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    meanDeltaPerc = meanDelta.div(theoreticalMean).mul(100);
    maxMeanDeltaPerc = meanDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxMeanDeltaPerc < tolerancePerc);

    stdDelta = Transforms.abs(theoreticalMean.sub(myNormalizer.getMean()));
    stdDeltaPerc = stdDelta.div(theoreticalStd).mul(100);
    maxStdDeltaPerc = stdDeltaPerc.max(1).getDouble(0, 0);
    assertTrue(maxStdDeltaPerc < tolerancePerc);
}
 

@Test
public void normalizationTests() {
    List<List<Writable>> data = new ArrayList<>();
    Schema.Builder builder = new Schema.Builder();
    int numColumns = 6;
    for (int i = 0; i < numColumns; i++)
        builder.addColumnDouble(String.valueOf(i));

    for (int i = 0; i < 5; i++) {
        List<Writable> record = new ArrayList<>(numColumns);
        data.add(record);
        for (int j = 0; j < numColumns; j++) {
            record.add(new DoubleWritable(1.0));
        }

    }

    INDArray arr = RecordConverter.toMatrix(data);

    Schema schema = builder.build();
    JavaRDD<List<Writable>> rdd = sc.parallelize(data);
    assertEquals(schema, DataFrames.fromStructType(DataFrames.fromSchema(schema)));
    assertEquals(rdd.collect(), DataFrames.toRecords(DataFrames.toDataFrame(schema, rdd)).getSecond().collect());

    DataRowsFacade dataFrame = DataFrames.toDataFrame(schema, rdd);
    dataFrame.get().show();
    Normalization.zeromeanUnitVariance(dataFrame).get().show();
    Normalization.normalize(dataFrame).get().show();

    //assert equivalent to the ndarray pre processing
    NormalizerStandardize standardScaler = new NormalizerStandardize();
    standardScaler.fit(new DataSet(arr.dup(), arr.dup()));
    INDArray standardScalered = arr.dup();
    standardScaler.transform(new DataSet(standardScalered, standardScalered));
    DataNormalization zeroToOne = new NormalizerMinMaxScaler();
    zeroToOne.fit(new DataSet(arr.dup(), arr.dup()));
    INDArray zeroToOnes = arr.dup();
    zeroToOne.transform(new DataSet(zeroToOnes, zeroToOnes));

    INDArray zeroMeanUnitVarianceDataFrame =
                    RecordConverter.toMatrix(Normalization.zeromeanUnitVariance(schema, rdd).collect());
    INDArray zeroMeanUnitVarianceDataFrameZeroToOne =
                    RecordConverter.toMatrix(Normalization.normalize(schema, rdd).collect());
    assertEquals(standardScalered, zeroMeanUnitVarianceDataFrame);
    assertTrue(zeroToOnes.equalsWithEps(zeroMeanUnitVarianceDataFrameZeroToOne, 1e-1));

}
 

@Test
public void testTransform() {
    /*Random dataset is generated such that
        AX + B where X is from a normal distribution with mean 0 and std 1
        The mean of above will be B and std A
        Obtained mean and std dev are compared to theoretical
        Transformed values should be the same as X with the same seed.
     */
    long randSeed = 12345;

    int nFeatures = 2;
    int nSamples = 6400;
    int bsize = 8;
    int a = 5;
    int b = 100;
    INDArray sampleMean, sampleStd, sampleMeanDelta, sampleStdDelta, delta, deltaPerc;
    double maxDeltaPerc, sampleMeanSEM;

    genRandomDataSet normData = new genRandomDataSet(nSamples, nFeatures, a, b, randSeed);
    DataSet genRandExpected = normData.theoreticalTransform;
    genRandomDataSet expectedData = new genRandomDataSet(nSamples, nFeatures, 1, 0, randSeed);
    genRandomDataSet beforeTransformData = new genRandomDataSet(nSamples, nFeatures, a, b, randSeed);

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    DataSetIterator normIterator = normData.getIter(bsize);
    DataSetIterator genRandExpectedIter = new TestDataSetIterator(genRandExpected, bsize);
    DataSetIterator expectedIterator = expectedData.getIter(bsize);
    DataSetIterator beforeTransformIterator = beforeTransformData.getIter(bsize);

    myNormalizer.fit(normIterator);

    double tolerancePerc = 0.10; //within 0.1%
    sampleMean = myNormalizer.getMean();
    sampleMeanDelta = Transforms.abs(sampleMean.sub(normData.theoreticalMean));
    assertTrue(sampleMeanDelta.mul(100).div(normData.theoreticalMean).max().getDouble(0) < tolerancePerc);
    //sanity check to see if it's within the theoretical standard error of mean
    sampleMeanSEM = sampleMeanDelta.div(normData.theoreticalSEM).max().getDouble(0);
    assertTrue(sampleMeanSEM < 2.6); //99% of the time it should be within this many SEMs

    tolerancePerc = 1; //within 1% - std dev value
    sampleStd = myNormalizer.getStd();
    sampleStdDelta = Transforms.abs(sampleStd.sub(normData.theoreticalStd));

    double actualmaxDiff = sampleStdDelta.div(normData.theoreticalStd).max().mul(100).getDouble(0);
    assertTrue(actualmaxDiff < tolerancePerc);

    tolerancePerc = 1; //within 1%
    normIterator.setPreProcessor(myNormalizer);
    while (normIterator.hasNext()) {
        INDArray before = beforeTransformIterator.next().getFeatures();
        INDArray origBefore = genRandExpectedIter.next().getFeatures();
        INDArray after = normIterator.next().getFeatures();
        INDArray expected = expectedIterator.next().getFeatures();
        delta = Transforms.abs(after.sub(expected));
        deltaPerc = delta.div(Transforms.abs(before.sub(expected)));
        deltaPerc.muli(100);
        maxDeltaPerc = deltaPerc.max(0, 1).getDouble(0);
        /*
        System.out.println("=== BEFORE ===");
        System.out.println(before);
        System.out.println("=== ORIG BEFORE ===");
        System.out.println(origBefore);
        System.out.println("=== AFTER ===");
        System.out.println(after);
        System.out.println("=== SHOULD BE ===");
        System.out.println(expected);
        System.out.println("% diff, "+ maxDeltaPerc);
        */
        assertTrue(maxDeltaPerc < tolerancePerc);
    }
}
 

@Test
public void testBruteForce3dMaskLabels() {

    NormalizerStandardize myNormalizer = new NormalizerStandardize();
    myNormalizer.fitLabel(true);
    NormalizerMinMaxScaler myMinMaxScaler = new NormalizerMinMaxScaler();
    myMinMaxScaler.fitLabel(true);

    //generating a dataset with consecutive numbers as feature values. Dataset also has masks
    int samples = 100;
    INDArray featureScale = Nd4j.create(new double[] {1, 2, 10}).reshape(3, 1);
    int timeStepsU = 5;
    Construct3dDataSet sampleU = new Construct3dDataSet(featureScale, timeStepsU, samples, 1);
    int timeStepsV = 3;
    Construct3dDataSet sampleV = new Construct3dDataSet(featureScale, timeStepsV, samples, sampleU.newOrigin);
    List<DataSet> dataSetList = new ArrayList<DataSet>();
    dataSetList.add(sampleU.sampleDataSet);
    dataSetList.add(sampleV.sampleDataSet);

    DataSet fullDataSetA = DataSet.merge(dataSetList);
    DataSet fullDataSetAA = fullDataSetA.copy();
    //This should be the same datasets as above without a mask
    Construct3dDataSet fullDataSetNoMask =
                    new Construct3dDataSet(featureScale, timeStepsU + timeStepsV, samples, 1);

    //preprocessors - label and feature values are the same
    myNormalizer.fit(fullDataSetA);
    assertEquals(myNormalizer.getMean().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMean.castTo(DataType.FLOAT));
    assertEquals(myNormalizer.getStd().castTo(DataType.FLOAT), fullDataSetNoMask.expectedStd.castTo(DataType.FLOAT));
    assertEquals(myNormalizer.getLabelMean().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMean.castTo(DataType.FLOAT));
    assertEquals(myNormalizer.getLabelStd().castTo(DataType.FLOAT), fullDataSetNoMask.expectedStd.castTo(DataType.FLOAT));

    myMinMaxScaler.fit(fullDataSetAA);
    assertEquals(myMinMaxScaler.getMin().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMin.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getMax().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMax.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getLabelMin().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMin.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getLabelMax().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMax.castTo(DataType.FLOAT));


    //Same Test with an Iterator, values should be close for std, exact for everything else
    DataSetIterator sampleIterA = new TestDataSetIterator(fullDataSetA, 5);
    DataSetIterator sampleIterB = new TestDataSetIterator(fullDataSetAA, 5);

    myNormalizer.fit(sampleIterA);
    assertEquals(myNormalizer.getMean().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMean.castTo(DataType.FLOAT));
    assertEquals(myNormalizer.getLabelMean().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMean.castTo(DataType.FLOAT));
    double diff1 = Transforms.abs(myNormalizer.getStd().div(fullDataSetNoMask.expectedStd).sub(1)).maxNumber().doubleValue();
    double diff2 = Transforms.abs(myNormalizer.getLabelStd().div(fullDataSetNoMask.expectedStd).sub(1)).maxNumber().doubleValue();
    assertTrue(diff1 < 0.01);
    assertTrue(diff2 < 0.01);

    myMinMaxScaler.fit(sampleIterB);
    assertEquals(myMinMaxScaler.getMin().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMin.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getMax().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMax.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getLabelMin().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMin.castTo(DataType.FLOAT));
    assertEquals(myMinMaxScaler.getLabelMax().castTo(DataType.FLOAT), fullDataSetNoMask.expectedMax.castTo(DataType.FLOAT));
}