Examples with Updater org.deeplearning4j.nn.api.Updater used on opensource projects

Example 16 with Updater

use of org.deeplearning4j.nn.api.Updater in project deeplearning4j by deeplearning4j.

the class TestUpdaters method testNoOpUpdater.

@Test
public void testNoOpUpdater() {
    Random r = new Random(12345L);
    double lr = 0.5;
    NeuralNetConfiguration conf = new NeuralNetConfiguration.Builder().learningRate(lr).layer(new DenseLayer.Builder().nIn(nIn).nOut(nOut).updater(org.deeplearning4j.nn.conf.Updater.NONE).build()).build();
    int numParams = conf.getLayer().initializer().numParams(conf);
    INDArray params = Nd4j.create(1, numParams);
    Layer layer = conf.getLayer().instantiate(conf, null, 0, params, true);
    Updater updater = UpdaterCreator.getUpdater(layer);
    for (int i = 0; i < weightGradient.length(); i++) weightGradient.putScalar(i, r.nextDouble());
    for (int i = 0; i < biasGradient.length(); i++) biasGradient.putScalar(i, r.nextDouble());
    gradient.gradientForVariable().put(DefaultParamInitializer.WEIGHT_KEY, weightGradient);
    gradient.gradientForVariable().put(DefaultParamInitializer.BIAS_KEY, biasGradient);
    updater.update(layer, gradient, -1, 1);
    INDArray weightGradActual = gradient.getGradientFor(DefaultParamInitializer.WEIGHT_KEY);
    INDArray biasGradActual = gradient.getGradientFor(DefaultParamInitializer.BIAS_KEY);
    assertEquals(weightGradient, weightGradActual);
    assertEquals(biasGradient, biasGradActual);
}

Example 17 with Updater

use of org.deeplearning4j.nn.api.Updater in project deeplearning4j by deeplearning4j.

the class VariationalAutoencoder method fit.

@Override
public void fit() {
    if (input == null) {
        throw new IllegalStateException("Cannot fit layer: layer input is null (not set)");
    }
    if (solver == null) {
        solver = new Solver.Builder().model(this).configure(conf()).listeners(getListeners()).build();
        //Set the updater state view array. For MLN and CG, this is done by MultiLayerUpdater and ComputationGraphUpdater respectively
        Updater updater = solver.getOptimizer().getUpdater();
        int updaterStateSize = updater.stateSizeForLayer(this);
        if (updaterStateSize > 0) {
            updater.setStateViewArray(this, Nd4j.createUninitialized(new int[] { 1, updaterStateSize }, Nd4j.order()), true);
        }
    }
    this.optimizer = solver.getOptimizer();
    solver.optimize();
}

Example 18 with Updater

use of org.deeplearning4j.nn.api.Updater in project deeplearning4j by deeplearning4j.

the class BaseOutputLayer method fit.

/**
     * Fit the model
     *
     * @param input the examples to classify (one example in each row)
     * @param labels   the example labels(a binary outcome matrix)
     */
@Override
public void fit(INDArray input, INDArray labels) {
    setInput(input);
    setLabels(labels);
    applyDropOutIfNecessary(true);
    if (solver == null) {
        solver = new Solver.Builder().configure(conf()).listeners(getListeners()).model(this).build();
        //Set the updater state view array. For MLN and CG, this is done by MultiLayerUpdater and ComputationGraphUpdater respectively
        Updater updater = solver.getOptimizer().getUpdater();
        int updaterStateSize = updater.stateSizeForLayer(this);
        if (updaterStateSize > 0)
            updater.setStateViewArray(this, Nd4j.createUninitialized(new int[] { 1, updaterStateSize }, Nd4j.order()), true);
    }
    solver.optimize();
}

Example 19 with Updater

use of org.deeplearning4j.nn.api.Updater in project deeplearning4j by deeplearning4j.

the class BaseLayer method fit.

@Override
public void fit(INDArray input) {
    if (input != null) {
        setInput(input.dup());
        applyDropOutIfNecessary(true);
    }
    if (solver == null) {
        solver = new Solver.Builder().model(this).configure(conf()).listeners(getListeners()).build();
        //Set the updater state view array. For MLN and CG, this is done by MultiLayerUpdater and ComputationGraphUpdater respectively
        Updater updater = solver.getOptimizer().getUpdater();
        int updaterStateSize = updater.stateSizeForLayer(this);
        if (updaterStateSize > 0)
            updater.setStateViewArray(this, Nd4j.createUninitialized(new int[] { 1, updaterStateSize }, Nd4j.order()), true);
    }
    this.optimizer = solver.getOptimizer();
    solver.optimize();
}

Example 20 with Updater

use of org.deeplearning4j.nn.api.Updater in project deeplearning4j by deeplearning4j.

the class TestDecayPolicies method testMomentumScheduleSingleLayer.

@Test
public void testMomentumScheduleSingleLayer() {
    double lr = 1e-2;
    double mu = 0.6;
    Map<Integer, Double> momentumAfter = new HashMap<>();
    momentumAfter.put(1, 0.2);
    int iterations = 2;
    NeuralNetConfiguration conf = new NeuralNetConfiguration.Builder().learningRate(lr).momentum(mu).momentumAfter(momentumAfter).iterations(iterations).layer(new DenseLayer.Builder().nIn(nIn).nOut(nOut).updater(org.deeplearning4j.nn.conf.Updater.NESTEROVS).build()).build();
    int numParams = conf.getLayer().initializer().numParams(conf);
    INDArray params = Nd4j.create(1, numParams);
    Layer layer = conf.getLayer().instantiate(conf, null, 0, params, true);
    Updater updater = UpdaterCreator.getUpdater(layer);
    int stateSize = updater.stateSizeForLayer(layer);
    updater.setStateViewArray(layer, Nd4j.create(1, stateSize), true);
    Gradient gradientExpected = new DefaultGradient();
    gradientExpected.setGradientFor(DefaultParamInitializer.WEIGHT_KEY, weightGradient.dup());
    gradientExpected.setGradientFor(DefaultParamInitializer.BIAS_KEY, biasGradient.dup());
    for (int i = 0; i < 2; i++) {
        updater.update(layer, gradientSingle, i, 1);
        mu = testNesterovsComputation(gradientSingle, gradientExpected, lr, mu, momentumAfter, i);
        assertEquals(mu, layer.conf().getLayer().getMomentum(), 1e-4);
    }
}