Examples with PreprocessorVertex org.deeplearning4j.nn.conf.graph.PreprocessorVertex used on opensource projects

Example 1 with PreprocessorVertex

use of org.deeplearning4j.nn.conf.graph.PreprocessorVertex in project deeplearning4j by deeplearning4j.

the class KerasModel method getComputationGraphConfiguration.

/**
     * Configure a ComputationGraph from this Keras Model configuration.
     *
     * @return          ComputationGraph
     */
public ComputationGraphConfiguration getComputationGraphConfiguration() throws InvalidKerasConfigurationException, UnsupportedKerasConfigurationException {
    if (!this.className.equals(MODEL_CLASS_NAME_MODEL) && !this.className.equals(MODEL_CLASS_NAME_SEQUENTIAL))
        throw new InvalidKerasConfigurationException("Keras model class name " + this.className + " incompatible with ComputationGraph");
    NeuralNetConfiguration.Builder modelBuilder = new NeuralNetConfiguration.Builder();
    ComputationGraphConfiguration.GraphBuilder graphBuilder = modelBuilder.graphBuilder();
    /* Build String array of input layer names, add to ComputationGraph. */
    String[] inputLayerNameArray = new String[this.inputLayerNames.size()];
    this.inputLayerNames.toArray(inputLayerNameArray);
    graphBuilder.addInputs(inputLayerNameArray);
    /* Build InputType array of input layer types, add to ComputationGraph. */
    List<InputType> inputTypeList = new ArrayList<InputType>();
    for (String inputLayerName : this.inputLayerNames) inputTypeList.add(this.layers.get(inputLayerName).getOutputType());
    InputType[] inputTypes = new InputType[inputTypeList.size()];
    inputTypeList.toArray(inputTypes);
    graphBuilder.setInputTypes(inputTypes);
    /* Build String array of output layer names, add to ComputationGraph. */
    String[] outputLayerNameArray = new String[this.outputLayerNames.size()];
    this.outputLayerNames.toArray(outputLayerNameArray);
    graphBuilder.setOutputs(outputLayerNameArray);
    Map<String, InputPreProcessor> preprocessors = new HashMap<String, InputPreProcessor>();
    /* Add layersOrdered one at a time. */
    for (KerasLayer layer : this.layersOrdered) {
        /* Get inbound layer names. */
        List<String> inboundLayerNames = layer.getInboundLayerNames();
        String[] inboundLayerNamesArray = new String[inboundLayerNames.size()];
        inboundLayerNames.toArray(inboundLayerNamesArray);
        /* Get inbound InputTypes and InputPreProcessor, if necessary. */
        List<InputType> inboundTypeList = new ArrayList<InputType>();
        for (String layerName : inboundLayerNames) inboundTypeList.add(this.outputTypes.get(layerName));
        InputType[] inboundTypeArray = new InputType[inboundTypeList.size()];
        inboundTypeList.toArray(inboundTypeArray);
        InputPreProcessor preprocessor = layer.getInputPreprocessor(inboundTypeArray);
        if (layer.usesRegularization())
            modelBuilder.setUseRegularization(true);
        if (layer.isLayer()) {
            /* Add DL4J layer. */
            if (preprocessor != null)
                preprocessors.put(layer.getLayerName(), preprocessor);
            graphBuilder.addLayer(layer.getLayerName(), layer.getLayer(), inboundLayerNamesArray);
            if (this.outputLayerNames.contains(layer.getLayerName()) && !(layer.getLayer() instanceof IOutputLayer))
                log.warn("Model cannot be trained: output layer " + layer.getLayerName() + " is not an IOutputLayer (no loss function specified)");
        } else if (layer.isVertex()) {
            /* Add DL4J vertex. */
            if (preprocessor != null)
                preprocessors.put(layer.getLayerName(), preprocessor);
            graphBuilder.addVertex(layer.getLayerName(), layer.getVertex(), inboundLayerNamesArray);
            if (this.outputLayerNames.contains(layer.getLayerName()) && !(layer.getVertex() instanceof IOutputLayer))
                log.warn("Model cannot be trained: output vertex " + layer.getLayerName() + " is not an IOutputLayer (no loss function specified)");
        } else if (layer.isInputPreProcessor()) {
            if (preprocessor == null)
                throw new UnsupportedKerasConfigurationException("Layer " + layer.getLayerName() + " could not be mapped to Layer, Vertex, or InputPreProcessor");
            graphBuilder.addVertex(layer.getLayerName(), new PreprocessorVertex(preprocessor), inboundLayerNamesArray);
        }
        if (this.outputLayerNames.contains(layer.getLayerName()))
            log.warn("Model cannot be trained: output " + layer.getLayerName() + " is not an IOutputLayer (no loss function specified)");
    }
    graphBuilder.setInputPreProcessors(preprocessors);
    /* Whether to use standard backprop (or BPTT) or truncated BPTT. */
    if (this.useTruncatedBPTT && this.truncatedBPTT > 0)
        graphBuilder.backpropType(BackpropType.TruncatedBPTT).tBPTTForwardLength(truncatedBPTT).tBPTTBackwardLength(truncatedBPTT);
    else
        graphBuilder.backpropType(BackpropType.Standard);
    return graphBuilder.build();
}

Example 2 with PreprocessorVertex

use of org.deeplearning4j.nn.conf.graph.PreprocessorVertex in project deeplearning4j by deeplearning4j.

the class TestGraphNodes method testJSON.

@Test
public void testJSON() {
    //The config here is non-sense, but that doesn't matter for config -> json -> config test
    ComputationGraphConfiguration conf = new NeuralNetConfiguration.Builder().graphBuilder().addInputs("in").addVertex("v1", new ElementWiseVertex(ElementWiseVertex.Op.Add), "in").addVertex("v2", new org.deeplearning4j.nn.conf.graph.MergeVertex(), "in", "in").addVertex("v3", new PreprocessorVertex(new CnnToFeedForwardPreProcessor(1, 2, 1)), "in").addVertex("v4", new org.deeplearning4j.nn.conf.graph.SubsetVertex(0, 1), "in").addVertex("v5", new DuplicateToTimeSeriesVertex("in"), "in").addVertex("v6", new LastTimeStepVertex("in"), "in").addVertex("v7", new org.deeplearning4j.nn.conf.graph.StackVertex(), "in").addVertex("v8", new org.deeplearning4j.nn.conf.graph.UnstackVertex(0, 1), "in").addLayer("out", new OutputLayer.Builder().nIn(1).nOut(1).build(), "in").setOutputs("out").build();
    String json = conf.toJson();
    ComputationGraphConfiguration conf2 = ComputationGraphConfiguration.fromJson(json);
    assertEquals(conf, conf2);
}

Example 3 with PreprocessorVertex

use of org.deeplearning4j.nn.conf.graph.PreprocessorVertex in project deeplearning4j by deeplearning4j.

the class FlowIterationListenerTest method setUp.

@Before
public void setUp() throws Exception {
    if (graph == null) {
        int VOCAB_SIZE = 1000;
        ComputationGraphConfiguration configuration = new NeuralNetConfiguration.Builder().regularization(true).l2(0.0001).weightInit(WeightInit.XAVIER).learningRate(0.01).updater(Updater.RMSPROP).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1).graphBuilder().addInputs("inEn", "inFr").setInputTypes(InputType.recurrent(VOCAB_SIZE + 1), InputType.recurrent(VOCAB_SIZE + 1)).addLayer("embeddingEn", new EmbeddingLayer.Builder().nIn(VOCAB_SIZE + 1).nOut(128).activation(Activation.IDENTITY).build(), "inEn").addLayer("encoder", new GravesLSTM.Builder().nIn(128).nOut(256).activation(Activation.SOFTSIGN).build(), "embeddingEn").addVertex("lastTimeStep", new LastTimeStepVertex("inEn"), "encoder").addVertex("duplicateTimeStep", new DuplicateToTimeSeriesVertex("inFr"), "lastTimeStep").addLayer("embeddingFr", new EmbeddingLayer.Builder().nIn(VOCAB_SIZE + 1).nOut(128).activation(Activation.IDENTITY).build(), "inFr").addVertex("embeddingFrSeq", new PreprocessorVertex(new FeedForwardToRnnPreProcessor()), "embeddingFr").addLayer("decoder", new GravesLSTM.Builder().nIn(128 + 256).nOut(256).activation(Activation.SOFTSIGN).build(), "embeddingFrSeq", "duplicateTimeStep").addLayer("output", new RnnOutputLayer.Builder().nIn(256).nOut(VOCAB_SIZE + 1).activation(Activation.SOFTMAX).build(), "decoder").setOutputs("output").pretrain(false).backprop(true).build();
        graph = new ComputationGraph(configuration);
        graph.init();
        INDArray input = Nd4j.zeros(10, VOCAB_SIZE, 20);
        graph.setInputs(input, input);
    }
    if (network == null) {
        final int numRows = 40;
        final int numColumns = 40;
        int nChannels = 3;
        int outputNum = LFWLoader.NUM_LABELS;
        int numSamples = LFWLoader.NUM_IMAGES;
        boolean useSubset = false;
        // numSamples/10;
        int batchSize = 200;
        int iterations = 5;
        int splitTrainNum = (int) (batchSize * .8);
        int seed = 123;
        int listenerFreq = iterations / 5;
        DataSet lfwNext;
        SplitTestAndTrain trainTest;
        DataSet trainInput;
        List<INDArray> testInput = new ArrayList<>();
        List<INDArray> testLabels = new ArrayList<>();
        MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().seed(seed).iterations(iterations).activation(Activation.RELU).weightInit(WeightInit.XAVIER).gradientNormalization(GradientNormalization.RenormalizeL2PerLayer).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).learningRate(0.01).momentum(0.9).regularization(true).updater(Updater.ADAGRAD).useDropConnect(true).list().layer(0, new ConvolutionLayer.Builder(4, 4).name("cnn1").nIn(nChannels).stride(1, 1).nOut(20).build()).layer(1, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX, new int[] { 2, 2 }).name("pool1").build()).layer(2, new ConvolutionLayer.Builder(3, 3).name("cnn2").stride(1, 1).nOut(40).build()).layer(3, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX, new int[] { 2, 2 }).name("pool2").build()).layer(4, new ConvolutionLayer.Builder(3, 3).name("cnn3").stride(1, 1).nOut(60).build()).layer(5, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX, new int[] { 2, 2 }).name("pool3").build()).layer(6, new ConvolutionLayer.Builder(2, 2).name("cnn4").stride(1, 1).nOut(80).build()).layer(7, new DenseLayer.Builder().name("ffn1").nOut(160).dropOut(0.5).build()).layer(8, new OutputLayer.Builder(LossFunctions.LossFunction.NEGATIVELOGLIKELIHOOD).nOut(outputNum).activation(Activation.SOFTMAX).build()).backprop(true).pretrain(false);
        new ConvolutionLayerSetup(builder, numRows, numColumns, nChannels);
        network = new MultiLayerNetwork(builder.build());
        network.init();
        INDArray input = Nd4j.zeros(10, nChannels, numRows, numColumns);
        network.setInput(input);
    }
}