Examples with DataSet org.nd4j.linalg.dataset.DataSet used on opensource projects

Example 26 with DataSet

use of org.nd4j.linalg.dataset.DataSet in project deeplearning4j by deeplearning4j.

the class TestDataVecDataSetFunctions method testDataVecSequencePairDataSetFunctionVariableLength.

@Test
public void testDataVecSequencePairDataSetFunctionVariableLength() throws Exception {
    //Same sort of test as testDataVecSequencePairDataSetFunction() but with variable length time series (labels shorter, align end)
    //Convert data to a SequenceFile:
    File f = new File("src/test/resources/csvsequence/csvsequence_0.txt");
    String pathFeatures = f.getAbsolutePath();
    String folderFeatures = pathFeatures.substring(0, pathFeatures.length() - 17);
    pathFeatures = folderFeatures + "*";
    File f2 = new File("src/test/resources/csvsequencelabels/csvsequencelabelsShort_0.txt");
    String pathLabels = f2.getPath();
    String folderLabels = pathLabels.substring(0, pathLabels.length() - 28);
    pathLabels = folderLabels + "*";
    //Extract a number from the file name
    PathToKeyConverter pathConverter = new PathToKeyConverterNumber();
    JavaPairRDD<Text, BytesPairWritable> toWrite = DataVecSparkUtil.combineFilesForSequenceFile(sc, pathFeatures, pathLabels, pathConverter);
    Path p = Files.createTempDirectory("dl4j_testSeqPairFnVarLength");
    p.toFile().deleteOnExit();
    String outPath = p.toString() + "/out";
    new File(outPath).deleteOnExit();
    toWrite.saveAsNewAPIHadoopFile(outPath, Text.class, BytesPairWritable.class, SequenceFileOutputFormat.class);
    //Load from sequence file:
    JavaPairRDD<Text, BytesPairWritable> fromSeq = sc.sequenceFile(outPath, Text.class, BytesPairWritable.class);
    SequenceRecordReader srr1 = new CSVSequenceRecordReader(1, ",");
    SequenceRecordReader srr2 = new CSVSequenceRecordReader(1, ",");
    PairSequenceRecordReaderBytesFunction psrbf = new PairSequenceRecordReaderBytesFunction(srr1, srr2);
    JavaRDD<Tuple2<List<List<Writable>>, List<List<Writable>>>> writables = fromSeq.map(psrbf);
    //Map to DataSet:
    DataVecSequencePairDataSetFunction pairFn = new DataVecSequencePairDataSetFunction(4, false, DataVecSequencePairDataSetFunction.AlignmentMode.ALIGN_END);
    JavaRDD<DataSet> data = writables.map(pairFn);
    List<DataSet> sparkData = data.collect();
    //Now: do the same thing locally (SequenceRecordReaderDataSetIterator) and compare
    String featuresPath = f.getPath().replaceAll("0", "%d");
    String labelsPath = f2.getPath().replaceAll("0", "%d");
    SequenceRecordReader featureReader = new CSVSequenceRecordReader(1, ",");
    SequenceRecordReader labelReader = new CSVSequenceRecordReader(1, ",");
    featureReader.initialize(new NumberedFileInputSplit(featuresPath, 0, 2));
    labelReader.initialize(new NumberedFileInputSplit(labelsPath, 0, 2));
    SequenceRecordReaderDataSetIterator iter = new SequenceRecordReaderDataSetIterator(featureReader, labelReader, 1, 4, false, SequenceRecordReaderDataSetIterator.AlignmentMode.ALIGN_END);
    List<DataSet> localData = new ArrayList<>(3);
    while (iter.hasNext()) localData.add(iter.next());
    assertEquals(3, sparkData.size());
    assertEquals(3, localData.size());
    //1 example, 3 values, 4 time steps
    int[] fShapeExp = new int[] { 1, 3, 4 };
    //1 example, 4 values/classes, 4 time steps (after padding)
    int[] lShapeExp = new int[] { 1, 4, 4 };
    for (int i = 0; i < 3; i++) {
        //Check shapes etc. data sets order may differ for spark vs. local
        DataSet dsSpark = sparkData.get(i);
        DataSet dsLocal = localData.get(i);
        //Expect mask array for labels
        assertNotNull(dsSpark.getLabelsMaskArray());
        INDArray fSpark = dsSpark.getFeatureMatrix();
        INDArray fLocal = dsLocal.getFeatureMatrix();
        INDArray lSpark = dsSpark.getLabels();
        INDArray lLocal = dsLocal.getLabels();
        assertArrayEquals(fShapeExp, fSpark.shape());
        assertArrayEquals(fShapeExp, fLocal.shape());
        assertArrayEquals(lShapeExp, lSpark.shape());
        assertArrayEquals(lShapeExp, lLocal.shape());
    }
    //Check that results are the same (order not withstanding)
    boolean[] found = new boolean[3];
    for (int i = 0; i < 3; i++) {
        int foundIndex = -1;
        DataSet ds = sparkData.get(i);
        for (int j = 0; j < 3; j++) {
            if (ds.equals(localData.get(j))) {
                if (foundIndex != -1)
                    //Already found this value -> suggests this spark value equals two or more of local version? (Shouldn't happen)
                    fail();
                foundIndex = j;
                if (found[foundIndex])
                    //One of the other spark values was equal to this one -> suggests duplicates in Spark list
                    fail();
                //mark this one as seen before
                found[foundIndex] = true;
            }
        }
    }
    int count = 0;
    for (boolean b : found) if (b)
        count++;
    //Expect all 3 and exactly 3 pairwise matches between spark and local versions
    assertEquals(3, count);
    //-------------------------------------------------
    //NOW: test same thing, but for align start...
    DataVecSequencePairDataSetFunction pairFnAlignStart = new DataVecSequencePairDataSetFunction(4, false, DataVecSequencePairDataSetFunction.AlignmentMode.ALIGN_START);
    JavaRDD<DataSet> rddDataAlignStart = writables.map(pairFnAlignStart);
    List<DataSet> sparkDataAlignStart = rddDataAlignStart.collect();
    //re-initialize to reset
    featureReader.initialize(new NumberedFileInputSplit(featuresPath, 0, 2));
    labelReader.initialize(new NumberedFileInputSplit(labelsPath, 0, 2));
    SequenceRecordReaderDataSetIterator iterAlignStart = new SequenceRecordReaderDataSetIterator(featureReader, labelReader, 1, 4, false, SequenceRecordReaderDataSetIterator.AlignmentMode.ALIGN_START);
    List<DataSet> localDataAlignStart = new ArrayList<>(3);
    while (iterAlignStart.hasNext()) localDataAlignStart.add(iterAlignStart.next());
    assertEquals(3, sparkDataAlignStart.size());
    assertEquals(3, localDataAlignStart.size());
    for (int i = 0; i < 3; i++) {
        //Check shapes etc. data sets order may differ for spark vs. local
        DataSet dsSpark = sparkDataAlignStart.get(i);
        DataSet dsLocal = localDataAlignStart.get(i);
        //Expect mask array for labels
        assertNotNull(dsSpark.getLabelsMaskArray());
        INDArray fSpark = dsSpark.getFeatureMatrix();
        INDArray fLocal = dsLocal.getFeatureMatrix();
        INDArray lSpark = dsSpark.getLabels();
        INDArray lLocal = dsLocal.getLabels();
        assertArrayEquals(fShapeExp, fSpark.shape());
        assertArrayEquals(fShapeExp, fLocal.shape());
        assertArrayEquals(lShapeExp, lSpark.shape());
        assertArrayEquals(lShapeExp, lLocal.shape());
    }
    //Check that results are the same (order not withstanding)
    found = new boolean[3];
    for (int i = 0; i < 3; i++) {
        int foundIndex = -1;
        DataSet ds = sparkData.get(i);
        for (int j = 0; j < 3; j++) {
            if (ds.equals(localData.get(j))) {
                if (foundIndex != -1)
                    //Already found this value -> suggests this spark value equals two or more of local version? (Shouldn't happen)
                    fail();
                foundIndex = j;
                if (found[foundIndex])
                    //One of the other spark values was equal to this one -> suggests duplicates in Spark list
                    fail();
                //mark this one as seen before
                found[foundIndex] = true;
            }
        }
    }
    count = 0;
    for (boolean b : found) if (b)
        count++;
    //Expect all 3 and exactly 3 pairwise matches between spark and local versions
    assertEquals(3, count);
}

Example 27 with DataSet

use of org.nd4j.linalg.dataset.DataSet in project deeplearning4j by deeplearning4j.

the class TestDataVecDataSetFunctions method testDataVecSequencePairDataSetFunction.

@Test
public void testDataVecSequencePairDataSetFunction() throws Exception {
    JavaSparkContext sc = getContext();
    //Convert data to a SequenceFile:
    File f = new File("src/test/resources/csvsequence/csvsequence_0.txt");
    String path = f.getPath();
    String folder = path.substring(0, path.length() - 17);
    path = folder + "*";
    PathToKeyConverter pathConverter = new PathToKeyConverterFilename();
    JavaPairRDD<Text, BytesPairWritable> toWrite = DataVecSparkUtil.combineFilesForSequenceFile(sc, path, path, pathConverter);
    Path p = Files.createTempDirectory("dl4j_testSeqPairFn");
    p.toFile().deleteOnExit();
    String outPath = p.toString() + "/out";
    new File(outPath).deleteOnExit();
    toWrite.saveAsNewAPIHadoopFile(outPath, Text.class, BytesPairWritable.class, SequenceFileOutputFormat.class);
    //Load from sequence file:
    JavaPairRDD<Text, BytesPairWritable> fromSeq = sc.sequenceFile(outPath, Text.class, BytesPairWritable.class);
    SequenceRecordReader srr1 = new CSVSequenceRecordReader(1, ",");
    SequenceRecordReader srr2 = new CSVSequenceRecordReader(1, ",");
    PairSequenceRecordReaderBytesFunction psrbf = new PairSequenceRecordReaderBytesFunction(srr1, srr2);
    JavaRDD<Tuple2<List<List<Writable>>, List<List<Writable>>>> writables = fromSeq.map(psrbf);
    //Map to DataSet:
    DataVecSequencePairDataSetFunction pairFn = new DataVecSequencePairDataSetFunction();
    JavaRDD<DataSet> data = writables.map(pairFn);
    List<DataSet> sparkData = data.collect();
    //Now: do the same thing locally (SequenceRecordReaderDataSetIterator) and compare
    String featuresPath = f.getAbsolutePath().replaceAll("0", "%d");
    SequenceRecordReader featureReader = new CSVSequenceRecordReader(1, ",");
    SequenceRecordReader labelReader = new CSVSequenceRecordReader(1, ",");
    featureReader.initialize(new NumberedFileInputSplit(featuresPath, 0, 2));
    labelReader.initialize(new NumberedFileInputSplit(featuresPath, 0, 2));
    SequenceRecordReaderDataSetIterator iter = new SequenceRecordReaderDataSetIterator(featureReader, labelReader, 1, -1, true);
    List<DataSet> localData = new ArrayList<>(3);
    while (iter.hasNext()) localData.add(iter.next());
    assertEquals(3, sparkData.size());
    assertEquals(3, localData.size());
    for (int i = 0; i < 3; i++) {
        //Check shapes etc. data sets order may differ for spark vs. local
        DataSet dsSpark = sparkData.get(i);
        DataSet dsLocal = localData.get(i);
        assertNull(dsSpark.getFeaturesMaskArray());
        assertNull(dsSpark.getLabelsMaskArray());
        INDArray fSpark = dsSpark.getFeatureMatrix();
        INDArray fLocal = dsLocal.getFeatureMatrix();
        INDArray lSpark = dsSpark.getLabels();
        INDArray lLocal = dsLocal.getLabels();
        //1 example, 3 values, 3 time steps
        int[] s = new int[] { 1, 3, 4 };
        assertArrayEquals(s, fSpark.shape());
        assertArrayEquals(s, fLocal.shape());
        assertArrayEquals(s, lSpark.shape());
        assertArrayEquals(s, lLocal.shape());
    }
    //Check that results are the same (order not withstanding)
    boolean[] found = new boolean[3];
    for (int i = 0; i < 3; i++) {
        int foundIndex = -1;
        DataSet ds = sparkData.get(i);
        for (int j = 0; j < 3; j++) {
            if (ds.equals(localData.get(j))) {
                if (foundIndex != -1)
                    //Already found this value -> suggests this spark value equals two or more of local version? (Shouldn't happen)
                    fail();
                foundIndex = j;
                if (found[foundIndex])
                    //One of the other spark values was equal to this one -> suggests duplicates in Spark list
                    fail();
                //mark this one as seen before
                found[foundIndex] = true;
            }
        }
    }
    int count = 0;
    for (boolean b : found) if (b)
        count++;
    //Expect all 3 and exactly 3 pairwise matches between spark and local versions
    assertEquals(3, count);
}

Example 28 with DataSet

use of org.nd4j.linalg.dataset.DataSet in project deeplearning4j by deeplearning4j.

the class BaseSparkTest method getBasicSparkDataSet.

protected JavaRDD<DataSet> getBasicSparkDataSet(int nRows, INDArray input, INDArray labels) {
    List<DataSet> list = new ArrayList<>();
    for (int i = 0; i < nRows; i++) {
        INDArray inRow = input.getRow(i).dup();
        INDArray outRow = labels.getRow(i).dup();
        DataSet ds = new DataSet(inRow, outRow);
        list.add(ds);
    }
    list.iterator();
    data = new DataSet().merge(list);
    return sc.parallelize(list);
}

Example 29 with DataSet

use of org.nd4j.linalg.dataset.DataSet in project deeplearning4j by deeplearning4j.

the class TestEarlyStoppingSpark method testNoImprovementNEpochsTermination.

@Test
public void testNoImprovementNEpochsTermination() {
    //Idea: terminate training if score (test set loss) does not improve for 5 consecutive epochs
    //Simulate this by setting LR = 0.0
    Nd4j.getRandom().setSeed(12345);
    MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder().seed(12345).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1).updater(Updater.SGD).learningRate(0.0).weightInit(WeightInit.XAVIER).list().layer(0, new OutputLayer.Builder().nIn(4).nOut(3).lossFunction(LossFunctions.LossFunction.MCXENT).build()).pretrain(false).backprop(true).build();
    MultiLayerNetwork net = new MultiLayerNetwork(conf);
    net.setListeners(new ScoreIterationListener(1));
    JavaRDD<DataSet> irisData = getIris();
    EarlyStoppingModelSaver<MultiLayerNetwork> saver = new InMemoryModelSaver<>();
    EarlyStoppingConfiguration<MultiLayerNetwork> esConf = new EarlyStoppingConfiguration.Builder<MultiLayerNetwork>().epochTerminationConditions(new MaxEpochsTerminationCondition(100), new ScoreImprovementEpochTerminationCondition(5)).iterationTerminationConditions(//Initial score is ~2.5
    new MaxScoreIterationTerminationCondition(7.5)).scoreCalculator(new SparkDataSetLossCalculator(irisData, true, sc.sc())).modelSaver(saver).build();
    IEarlyStoppingTrainer<MultiLayerNetwork> trainer = new SparkEarlyStoppingTrainer(getContext().sc(), new ParameterAveragingTrainingMaster(true, 4, 1, 150 / 10, 1, 0), esConf, net, irisData);
    EarlyStoppingResult result = trainer.fit();
    //Expect no score change due to 0 LR -> terminate after 6 total epochs
    //Normally expect 6 epochs exactly; get a little more than that here due to rounding + order of operations
    assertTrue(result.getTotalEpochs() < 12);
    assertEquals(EarlyStoppingResult.TerminationReason.EpochTerminationCondition, result.getTerminationReason());
    String expDetails = new ScoreImprovementEpochTerminationCondition(5).toString();
    assertEquals(expDetails, result.getTerminationDetails());
}

Example 30 with DataSet

use of org.nd4j.linalg.dataset.DataSet in project deeplearning4j by deeplearning4j.

the class TestEarlyStoppingSpark method getIris.

private JavaRDD<DataSet> getIris() {
    JavaSparkContext sc = getContext();
    IrisDataSetIterator iter = new IrisDataSetIterator(irisBatchSize(), 150);
    List<DataSet> list = new ArrayList<>(150);
    while (iter.hasNext()) list.add(iter.next());
    return sc.parallelize(list);
}