Examples with ScoreIterationListener org.deeplearning4j.optimize.listeners.ScoreIterationListener used on opensource projects

Example 11 with ScoreIterationListener

use of org.deeplearning4j.optimize.listeners.ScoreIterationListener 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 12 with ScoreIterationListener

use of org.deeplearning4j.optimize.listeners.ScoreIterationListener in project deeplearning4j by deeplearning4j.

the class TestEarlyStoppingSpark method testTimeTermination.

@Test
public void testTimeTermination() {
    //test termination after max time
    Nd4j.getRandom().setSeed(12345);
    MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder().seed(12345).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1).updater(Updater.SGD).learningRate(1e-6).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(10000)).iterationTerminationConditions(new MaxTimeIterationTerminationCondition(3, TimeUnit.SECONDS), //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 / 15, 1, 0), esConf, net, irisData);
    long startTime = System.currentTimeMillis();
    EarlyStoppingResult result = trainer.fit();
    long endTime = System.currentTimeMillis();
    int durationSeconds = (int) (endTime - startTime) / 1000;
    assertTrue("durationSeconds = " + durationSeconds, durationSeconds >= 3);
    assertTrue("durationSeconds = " + durationSeconds, durationSeconds <= 9);
    assertEquals(EarlyStoppingResult.TerminationReason.IterationTerminationCondition, result.getTerminationReason());
    String expDetails = new MaxTimeIterationTerminationCondition(3, TimeUnit.SECONDS).toString();
    assertEquals(expDetails, result.getTerminationDetails());
}

Example 13 with ScoreIterationListener

use of org.deeplearning4j.optimize.listeners.ScoreIterationListener in project deeplearning4j by deeplearning4j.

the class ParallelWrapperTest method testParallelWrapperRun.

@Test
public void testParallelWrapperRun() throws Exception {
    int nChannels = 1;
    int outputNum = 10;
    // for GPU you usually want to have higher batchSize
    int batchSize = 128;
    int nEpochs = 10;
    int iterations = 1;
    int seed = 123;
    log.info("Load data....");
    DataSetIterator mnistTrain = new MnistDataSetIterator(batchSize, true, 12345);
    DataSetIterator mnistTest = new MnistDataSetIterator(batchSize, false, 12345);
    log.info("Build model....");
    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().seed(seed).iterations(iterations).regularization(true).l2(0.0005).learningRate(//.biasLearningRate(0.02)
    0.01).weightInit(WeightInit.XAVIER).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).updater(Updater.NESTEROVS).momentum(0.9).list().layer(0, new ConvolutionLayer.Builder(5, 5).nIn(nChannels).stride(1, 1).nOut(20).activation(Activation.IDENTITY).build()).layer(1, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX).kernelSize(2, 2).stride(2, 2).build()).layer(2, new ConvolutionLayer.Builder(5, 5).stride(1, 1).nOut(50).activation(Activation.IDENTITY).build()).layer(3, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX).kernelSize(2, 2).stride(2, 2).build()).layer(4, new DenseLayer.Builder().activation(Activation.RELU).nOut(500).build()).layer(5, new OutputLayer.Builder(LossFunctions.LossFunction.NEGATIVELOGLIKELIHOOD).nOut(outputNum).activation(Activation.SOFTMAX).build()).backprop(true).pretrain(false);
    // The builder needs the dimensions of the image along with the number of channels. these are 28x28 images in one channel
    new ConvolutionLayerSetup(builder, 28, 28, 1);
    MultiLayerConfiguration conf = builder.build();
    MultiLayerNetwork model = new MultiLayerNetwork(conf);
    model.init();
    // ParallelWrapper will take care of load balancing between GPUs.
    ParallelWrapper wrapper = new ParallelWrapper.Builder(model).prefetchBuffer(24).workers(2).averagingFrequency(3).reportScoreAfterAveraging(true).useLegacyAveraging(true).build();
    log.info("Train model....");
    model.setListeners(new ScoreIterationListener(100));
    long timeX = System.currentTimeMillis();
    for (int i = 0; i < nEpochs; i++) {
        long time1 = System.currentTimeMillis();
        // Please note: we're feeding ParallelWrapper with iterator, not model directly
        //            wrapper.fit(mnistMultiEpochIterator);
        wrapper.fit(mnistTrain);
        long time2 = System.currentTimeMillis();
        log.info("*** Completed epoch {}, time: {} ***", i, (time2 - time1));
    }
    long timeY = System.currentTimeMillis();
    log.info("*** Training complete, time: {} ***", (timeY - timeX));
    log.info("Evaluate model....");
    Evaluation eval = new Evaluation(outputNum);
    while (mnistTest.hasNext()) {
        DataSet ds = mnistTest.next();
        INDArray output = model.output(ds.getFeatureMatrix(), false);
        eval.eval(ds.getLabels(), output);
    }
    log.info(eval.stats());
    mnistTest.reset();
    log.info("****************Example finished********************");
    wrapper.shutdown();
}

Example 14 with ScoreIterationListener

use of org.deeplearning4j.optimize.listeners.ScoreIterationListener in project deeplearning4j by deeplearning4j.

the class TestParallelEarlyStopping method testBadTuning.

@Test
public void testBadTuning() {
    //Test poor tuning (high LR): should terminate on MaxScoreIterationTerminationCondition
    Nd4j.getRandom().setSeed(12345);
    MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder().seed(12345).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1).updater(Updater.SGD).learningRate(//Intentionally huge LR
    1.0).weightInit(WeightInit.XAVIER).list().layer(0, new OutputLayer.Builder().nIn(4).nOut(3).activation(Activation.SOFTMAX).lossFunction(LossFunctions.LossFunction.MCXENT).build()).pretrain(false).backprop(true).build();
    MultiLayerNetwork net = new MultiLayerNetwork(conf);
    net.setListeners(new ScoreIterationListener(1));
    DataSetIterator irisIter = new IrisDataSetIterator(10, 150);
    EarlyStoppingModelSaver<MultiLayerNetwork> saver = new InMemoryModelSaver<>();
    EarlyStoppingConfiguration<MultiLayerNetwork> esConf = new EarlyStoppingConfiguration.Builder<MultiLayerNetwork>().epochTerminationConditions(new MaxEpochsTerminationCondition(5000)).iterationTerminationConditions(new MaxTimeIterationTerminationCondition(1, TimeUnit.MINUTES), //Initial score is ~2.5
    new MaxScoreIterationTerminationCondition(10)).scoreCalculator(new DataSetLossCalculator(irisIter, true)).modelSaver(saver).build();
    IEarlyStoppingTrainer<MultiLayerNetwork> trainer = new EarlyStoppingParallelTrainer<>(esConf, net, irisIter, null, 2, 2, 1);
    EarlyStoppingResult result = trainer.fit();
    assertTrue(result.getTotalEpochs() < 5);
    assertEquals(EarlyStoppingResult.TerminationReason.IterationTerminationCondition, result.getTerminationReason());
    String expDetails = new MaxScoreIterationTerminationCondition(10).toString();
    assertEquals(expDetails, result.getTerminationDetails());
    assertTrue(result.getBestModelEpoch() <= 0);
    assertNotNull(result.getBestModel());
}

Example 15 with ScoreIterationListener

use of org.deeplearning4j.optimize.listeners.ScoreIterationListener in project deeplearning4j by deeplearning4j.

the class TestParallelEarlyStopping method testEarlyStoppingEveryNEpoch.

// parallel training results vary wildly with expected result
// need to determine if this test is feasible, and how it should
// be properly designed
//    @Test
//    public void testEarlyStoppingIris(){
//        MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder()
//                .optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1)
//                .updater(Updater.SGD)
//                .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));
//
//        DataSetIterator irisIter = new IrisDataSetIterator(50,600);
//        EarlyStoppingModelSaver<MultiLayerNetwork> saver = new InMemoryModelSaver<>();
//        EarlyStoppingConfiguration<MultiLayerNetwork> esConf = new EarlyStoppingConfiguration.Builder<MultiLayerNetwork>()
//            .epochTerminationConditions(new MaxEpochsTerminationCondition(5))
//            .evaluateEveryNEpochs(1)
//            .iterationTerminationConditions(new MaxTimeIterationTerminationCondition(1, TimeUnit.MINUTES))
//            .scoreCalculator(new DataSetLossCalculator(irisIter,true))
//            .modelSaver(saver)
//            .build();
//
//        IEarlyStoppingTrainer<MultiLayerNetwork> trainer = new EarlyStoppingParallelTrainer<>(esConf,net,irisIter,null,2,2,1);
//
//        EarlyStoppingResult<MultiLayerNetwork> result = trainer.fit();
//        System.out.println(result);
//
//        assertEquals(5, result.getTotalEpochs());
//        assertEquals(EarlyStoppingResult.TerminationReason.EpochTerminationCondition,result.getTerminationReason());
//        Map<Integer,Double> scoreVsIter = result.getScoreVsEpoch();
//        assertEquals(5,scoreVsIter.size());
//        String expDetails = esConf.getEpochTerminationConditions().get(0).toString();
//        assertEquals(expDetails, result.getTerminationDetails());
//
//        MultiLayerNetwork out = result.getBestModel();
//        assertNotNull(out);
//
//        //Check that best score actually matches (returned model vs. manually calculated score)
//        MultiLayerNetwork bestNetwork = result.getBestModel();
//        irisIter.reset();
//        double score = bestNetwork.score(irisIter.next());
//        assertEquals(result.getBestModelScore(), score, 1e-4);
//    }
@Test
public void testEarlyStoppingEveryNEpoch() {
    MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder().optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1).updater(Updater.SGD).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));
    DataSetIterator irisIter = new IrisDataSetIterator(50, 600);
    EarlyStoppingModelSaver<MultiLayerNetwork> saver = new InMemoryModelSaver<>();
    EarlyStoppingConfiguration<MultiLayerNetwork> esConf = new EarlyStoppingConfiguration.Builder<MultiLayerNetwork>().epochTerminationConditions(new MaxEpochsTerminationCondition(5)).scoreCalculator(new DataSetLossCalculator(irisIter, true)).evaluateEveryNEpochs(2).modelSaver(saver).build();
    IEarlyStoppingTrainer<MultiLayerNetwork> trainer = new EarlyStoppingParallelTrainer<>(esConf, net, irisIter, null, 2, 6, 1);
    EarlyStoppingResult<MultiLayerNetwork> result = trainer.fit();
    System.out.println(result);
    assertEquals(5, result.getTotalEpochs());
    assertEquals(EarlyStoppingResult.TerminationReason.EpochTerminationCondition, result.getTerminationReason());
}